# AI Tokenomics: What AI tokens cost and where they're wasted

> Source: <https://github.com/QuesmaOrg/awesome-ai-tokenomics>
> Published: 2026-07-26 11:39:19+00:00

A map of what AI tokens actually cost, and where they're wasted vs. well spent.

*"I feel nervous when I have subscription left over. That just means I haven't maximized my token throughput."*

Andrej Karpathy, [No Priors](https://podscripts.co/podcasts/no-priors-artificial-intelligence-technology-startups/andrej-karpathy-on-code-agents-autoresearch-and-the-loopy-era-of-ai) (2026)

As of 2026-07: 197 entries across five areas, 8 practices, 10 concepts, 8 claims, and 7 copy-paste setups.

**Topics:** [Caching](#caching) · [Compression](#compression) · [Context engineering](#context-engineering) · [Memory](#memory) · [Routing](#routing-model-selection) · [Multi-agent systems](#multi-agent-systems) · [Gateways](#gateways-and-proxies) · [Observability](#observability) · [Benchmarks](#benchmarks-evals) · [Cache accounting](#cache-accounting) · [Budgets](#budgets-caps) · [Pricing models](#pricing-models) · [Energy](#energy-carbon)

A hybrid by design: a curated awesome list at the core, plus a field-guide layer on top (practices, concepts, claims) that carries an editorial point of view.

[Where to start](#where-to-start)[Legend](#legend)[Monitor](#monitor)[Optimize](#optimize)[Govern](#govern)[Understand](#understand)[Measure](#measure)[Practices](#practices)[Concepts](#concepts)[Claims](#claims)[Setups and skills](#setups-and-skills)

Just want the numbers: the five area sections below hold every entry. Want the method: read the practices first, then the concepts behind them. Building something: setups and skills holds runnable configurations.

Each entry ends with a kind badge: [ (blue, with the license when known), or a gray badge for ](https://camo.githubusercontent.com/86b7430e00e49604cfd07fa3c23da1058217954821acd8ec53bcd42c7e321c95/68747470733a2f2f696d672e736869656c64732e696f2f62616467652f746f6f6c2d4d49542d626c75653f7374796c653d666c61742d737175617265)[, ](https://camo.githubusercontent.com/7cd52898962dcd7ff5471fdf312cf9337d8ab3047f40ddf70a046b4da7beb0d2/68747470733a2f2f696d672e736869656c64732e696f2f62616467652f70617065722d3535353f7374796c653d666c61742d737175617265)[, ](https://camo.githubusercontent.com/40e5f29fa348fa217d71539190e9868e4d0ad9d4c9110bdaa882f8b0ec1f917d/68747470733a2f2f696d672e736869656c64732e696f2f62616467652f62656e63682d3535353f7374796c653d666c61742d737175617265)[, ](https://camo.githubusercontent.com/86d64750f01441c070c94233c07c13d6a95cbc2d99b70d19dcae3432a515bb30/68747470733a2f2f696d672e736869656c64732e696f2f62616467652f646174612d3535353f7374796c653d666c61742d737175617265)[ for companies, and ](https://camo.githubusercontent.com/a5e1dc5c1c6980b67c5c73bc888acbb84d328fa26a7c624fc4fac91dfda60d30/68747470733a2f2f696d672e736869656c64732e696f2f62616467652f636f2d3535353f7374796c653d666c61742d737175617265)[. Plain entries are articles. GitHub-hosted tools also carry live star and last-commit badges.](https://camo.githubusercontent.com/ec4cf716c35b30b16f62e76c9bc115470f55d8116ce006af98ae4171f6e2c0bc/68747470733a2f2f696d672e736869656c64732e696f2f62616467652f7265706f72742d3535353f7374796c653d666c61742d737175617265)

[ccusage](https://github.com/ccusage/ccusage)- An open-source CLI that reads local agent logs to report token usage and cost across 15 coding-agent sources, with caching-aware pricing.[Claude Code Usage Monitor](https://github.com/Maciek-roboblog/Claude-Code-Usage-Monitor)- A live terminal dashboard for Claude Code usage, with burn-rate analytics, P90 limit detection, and session-expiry forecasts.[claude-usage](https://github.com/phuryn/claude-usage)- A local dashboard for Claude Code token usage, costs, and session history; Pro and Max subscribers get a quota progress bar.[ClaudeBar](https://github.com/tddworks/ClaudeBar)- A macOS menu-bar app that monitors AI coding quotas across 11 providers; the README declares MIT but ships no license file, so the OSS grant is unconfirmed.[CodeBurn](https://github.com/getagentseal/codeburn#find-and-fix-waste)- An open-source tracker for 36 coding tools whose optimize command flags named harness-waste patterns with dollar estimates it later checks against actuals.[Codex Usage Tracker](https://github.com/douglasmonsky/codex-usage-tracker)- A local-first dashboard, CLI, and MCP tools indexing Codex CLI logs into SQLite to show where tokens, credits, and cost go, including cache ratios.[CodexBar](https://github.com/steipete/CodexBar)- A free, open-source macOS menu-bar app that shows limits and reset timers at a glance across dozens of AI providers, plus credit balances and spending when…[CodeZeno Usage Monitor](https://github.com/CodeZeno/Claude-Code-Usage-Monitor)- A Windows-taskbar widget showing real-time Claude Code quota and usage at a glance, without opening a terminal.[Datadog LLM Observability - Cost](https://docs.datadoghq.com/llm_observability/monitoring/cost/)- Datadog's LLM Observability estimates per-request cost across 800+ models from token counts and public pricing; invoice reconciliation is a separate product.[gh-aw (GitHub Agentic Workflows)](https://github.com/github/gh-aw)- GitHub's agentic-workflows runtime with first-party per-run token and cost metering, plus budget caps that stop a workflow mid-run.[Grafana Cloud GenAI Observability](https://grafana.com/docs/grafana-cloud/monitor-applications/ai-observability/genai/observability/)- Grafana Cloud's GenAI Observability ships a prebuilt dashboard for LLM cost, token usage, and latency, built on top of the OpenLIT SDK.[OpenLIT](https://github.com/openlit/openlit)- An open-source (Apache-2.0), OpenTelemetry-native platform with a self-hosted dashboard for LLM cost, token, and latency observability.[OpenUsage](https://github.com/robinebers/openusage)- A native Swift macOS menu-bar meter for 10 AI coding subscriptions, showing session and weekly limits, credits, and estimated spend from local credentials.[TokenTracker](https://github.com/mm7894215/TokenTracker)- A local-first token and cost dashboard for 27 coding tools, with a desktop pet, native widgets, and achievements as a distinct gamified take on usage metering.

[AgentSight](https://github.com/eunomia-bpf/agentsight)- Uses eBPF to watch an AI agent from the kernel boundary, correlating what it said it would do against what it did, at under 3% overhead.[OpenTelemetry eBPF Instrumentation (OBI) - GenAI / MCP](https://opentelemetry.io/docs/zero-code/obi/)- OBI is OpenTelemetry's zero-code eBPF instrumentation (formerly Grafana Beyla) that captures GenAI and MCP traces at the kernel layer with no SDK.

[Langfuse](https://langfuse.com)- An open-source platform for tracing, evaluating, and analyzing LLM and agent transcripts, with a prompt-management layer on top.

[OpenLLMetry](https://github.com/traceloop/openllmetry)- An open-source set of OpenTelemetry-based SDKs and instrumentations, built by Traceloop, for LLM apps.[OpenTelemetry GenAI Semantic Conventions](https://github.com/open-telemetry/semantic-conventions-genai)- OpenTelemetry's GenAI Semantic Conventions define the vendor-neutral token, cost, and cache attribute names that OpenLLMetry and Phoenix both converge onto.

[Arize Phoenix](https://github.com/Arize-ai/phoenix)- A source-available (Elastic License 2.0) LLM tracing platform recording per-span token counts and USD cost via OpenTelemetry.[claude-tap](https://github.com/liaohch3/claude-tap)- A local trace viewer intercepting API traffic from 14+ coding agents, showing per-request token breakdowns: input, output, cache read, cache creation.[LangSmith - Cost Tracking](https://docs.langchain.com/langsmith/cost-tracking)- LangSmith is LangChain's commercial LLM/agent observability SaaS.[Opik](https://github.com/comet-ml/opik)- Comet's open-source (Apache-2.0) LLM observability platform, with per-span USD cost estimated from token usage.

[GPTCache](https://github.com/zilliztech/GPTCache)- An open-source semantic cache returning a stored LLM response for a paraphrased repeat query via vector search, skipping the paid call.[khazad](https://github.com/GuglielmoCerri/khazad)- A transport-layer semantic cache for LLM APIs on Redis 8 Vector Sets: it intercepts HTTP traffic with zero application code changes and replays cached responses…[LMCache - KV-cache reuse layer with token-level cache-hit accounting (self-host)](https://github.com/LMCache/LMCache)- LMCache is a self-hosted KV-cache layer beneath vLLM, giving token-level cache-hit observability for teams who own their GPUs, not a hosted bill.[prompt-cache](https://github.com/messkan/prompt-cache)- A Go LLM proxy that adds a three-tier semantic cache: high similarity hits directly, low skips, and a gray zone runs a cheap verification model to guard against…[Redis LangCache](https://redis.io/langcache/)- Redis's fully-managed semantic cache: a REST API that returns a stored response when a new query is similar to a past one, so paraphrases skip the paid LLM…

[llama.cpp](https://github.com/ggml-org/llama.cpp)- The foundational open-source (MIT) local LLM inference engine most of the local ecosystem runs on, with an OpenAI-compatible server built in.[Ollama](https://github.com/ollama/ollama)- A runtime for running open-weight models like Qwen, DeepSeek, and GLM-5.1 locally, shifting inference onto hardware you already own.

[Context Mode - MCP server that sandboxes tool output out of the context window](https://github.com/mksglu/context-mode)- This MCP server sandboxes tool calls and returns only the distilled result, claiming a 98% cut: 315 KB of output down to 5.4 KB.[headroom - context compression for LLM and agent pipelines (quality story dissected)](https://github.com/headroomlabs-ai/headroom)- An Apache-2.0 context-compression tool for LLM/agent pipelines, the category's largest repo, confirmed organic by star-forensics.[LLMLingua](https://github.com/microsoft/LLMLingua)- Microsoft's prompt-compression library that uses a small model to drop low-information tokens before a prompt reaches the target LLM.[llmtrim](https://github.com/fkiene/llmtrim)- A local proxy that compresses a coding agent's prompt, tool schemas, and history before forwarding, and can reroute Claude calls to Grok.[Minification of state-in-context agents - the clean waste-vs-capability datapoint](https://arxiv.org/abs/2606.01326)- This ICPC 2026 study found that minifying code in a coding agent's context cuts input tokens by 42% but costs 12 percentage points of accuracy.[rtk - CLI proxy that compresses dev-command output (reduction-only headline)](https://github.com/rtk-ai/rtk)- rtk is a single-binary Rust CLI proxy that intercepts and compresses the output of common dev commands before it reaches an LLM coding agent's context window.[TOON (Token-Oriented Object Notation)](https://github.com/toon-format/toon)- TOON is a compact, human-readable, lossless serialization of the JSON data model, designed for LLM input.

[AgentDiet - trajectory reduction ("Reducing Cost of LLM Agents with Trajectory Reduction")](https://arxiv.org/abs/2509.23586)- AgentDiet is an inference-time module that strips useless, redundant, and expired information from an agent's trajectory, without hurting performance.[Anthropic vendor-native context management (context editing + memory tool + server-side compaction)](https://platform.claude.com/docs/en/build-with-claude/context-editing)- Anthropic's context editing, memory tool, and server-side compaction cut token consumption by a vendor-reported 84% in a 100-turn web-search evaluation.[Claude Code compaction engine - the three-tier mechanism and its cache/correctness failure modes](https://barazany.dev/blog/claude-codes-compaction-engine)- Claude Code's*harness*- not the API - decides how to trim a filling context window, and it does this through a three-tier compaction engine.[Codex CLI compaction cost - over-eager compaction as a token-amplification loop](https://github.com/openai/codex/issues/16812)- Upgrading Codex CLI from v0.116 to v0.118 made context compaction fire twice as often, doubling or tripling token consumption for identical tasks.[Context Rot - LLM performance degrades as input length grows](https://www.trychroma.com/research/context-rot)- This is Chroma's controlled study of how LLM output quality changes as input length grows, holding task difficulty fixed.[ContextBudget - context management as a budget-constrained sequential decision](https://arxiv.org/abs/2604.01664)- ContextBudget's BACM method has an agent decide when and how much to compress its history based on remaining context budget, not a fixed rule.[Cursor vendor-native context management - dynamic context discovery + Composer self-summarization](https://cursor.com/blog/dynamic-context-discovery)- Cursor's dynamic context discovery loads tool schemas and large outputs on demand instead of eagerly, a change the vendor reports cut context usage by 46.9%.[Repomix](https://github.com/yamadashy/repomix)- Packs an entire repository into a single AI-friendly file, reporting token counts and using Tree-sitter to compress code to signatures only.[RULER - the "real context size" long-context benchmark](https://github.com/NVIDIA/RULER)- NVIDIA's RULER benchmark found that of models claiming 32K+ token context windows, only half actually maintain quality once you fill them to 32K.[Self-Compacting Language Model Agents](https://arxiv.org/abs/2606.23525)- This paper introduces SELFCOMPACT: instead of fixed-interval summarization, the model itself decides when and how to compress a growing agent trace.[Serena](https://github.com/oraios/serena)- An open-source (MIT) MCP toolkit that gives a coding agent IDE-grade semantic code retrieval and editing: 'the IDE for your coding agent'.

[Claude Code spend-governance bundle (v2.1.216-219) - caps tightened, fan-out default loosened](https://code.claude.com/docs/en/changelog)- Claude Code's v2.1.216-219 bundle (2026-07-20 to 07-24) hardens spend controls - a concurrent-subagent cap and enforced --max-budget-usd - while simultaneously…[Harness-side runaway-loop cost guardrails (Claude Code + Codex, July 2026)](https://github.com/anthropics/claude-code/releases/tag/v2.1.212)- In mid-July 2026 Claude Code and Codex both shipped first-party guardrails against runaway agent loops within days of each other.

[Bifrost (Maxim AI)](https://github.com/maximhq/bifrost)- Bifrost is a Go-based AI gateway fronting 1000+ models that measured just 11 microseconds of added latency per request at 5,000 requests per second.[Cloudflare AI Gateway (Spend Limits)](https://developers.cloudflare.com/ai-gateway/features/spend-limits/)- Cloudflare AI Gateway is an edge-native LLM proxy that added dollar-denominated spend limits in June 2026, blocking or rerouting requests once a budget is hit.[Helicone](https://helicone.ai)- An open-source (Apache-2.0) LLM proxy that logs every request's cost, latency, and tokens in one line of code; Mintlify acquired it in March 2026.[Kong AI Gateway](https://developer.konghq.com/ai-gateway/)- The AI layer of Kong's API-gateway platform: a proxy that meters LLM/agent/MCP traffic for billing, showback, and chargeback.[LiteLLM](https://github.com/BerriAI/litellm)- An open-source gateway fronting 100+ LLM APIs that computes real per-request dollar cost from a live pricing map, with spend limits.[OpenRouter](https://openrouter.ai/docs/guides/routing/provider-selection)- A unified API gateway fronting 400+ models across 70+ providers that auto-routes each request by price, with fallback on outages.[Portkey AI Gateway](https://portkey.ai/docs/product/ai-gateway/virtual-keys/budget-limits)- Routes LLM traffic across providers and enforces hard USD budget limits on virtual keys, auto-expiring a key once its cap is hit.

[claude-code-memory-setup](https://github.com/lucasrosati/claude-code-memory-setup)- A practitioner recipe pairing an Obsidian memory vault with a local AST code-graph tool; the author self-reports up to 71.5x fewer tokens per Claude Code…[claude-mem](https://github.com/thedotmack/claude-mem)- A coding-agent observational-memory layer that captures every session, compresses it with AI, and re-injects relevant context next time; self-reports ~10x token…[Cognee](https://github.com/topoteretes/cognee)- An open-source (Apache-2.0) AI-memory platform giving agents persistent memory via a self-hosted knowledge graph, via remember/recall/forget.[Karpathy's LLM Wiki](https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f)- Andrej Karpathy's LLM Wiki pattern has an agent build and maintain a persistent markdown wiki from your sources, instead of re-retrieving raw files.[LangMem](https://github.com/langchain-ai/langmem)- LangChain's long-term memory library: it extracts and consolidates facts from conversations and integrates natively with LangGraph's memory store, so agents…[Letta (MemGPT)](https://github.com/letta-ai/letta)- The MemGPT lineage project: a platform for stateful agents that pages an LLM's context like an OS, keeping working memory small and moving the rest to archival…[Mem0](https://github.com/mem0ai/mem0)- An open-source memory layer that extracts salient facts from conversations and retrieves only the relevant ones per call, not the full history.[Supermemory](https://github.com/supermemoryai/supermemory)- A memory and context engine that self-reports 95% recall on LongMemEval while adding only ~720 tokens of context, a claimed 99.4% context reduction versus full…[Zep / Graphiti](https://www.getzep.com/)- A memory platform for agents built on temporal knowledge graphs; it self-reports serving benchmark answers from a few thousand tokens of retrieved context… (also:[Graphiti (OSS engine)](https://github.com/getzep/graphiti)·[zep repo](https://github.com/getzep/zep))

[Framework orchestration overhead (the manager-LLM tax)](https://docs.crewai.com/en/learn/hierarchical-process)- Hierarchical frameworks like CrewAI add a manager-LLM delegation tax, an extra model that plans and validates, though its cost is unquantified in any primary.[SupervisorAgent - "Stop Wasting Your Tokens" (runtime supervision)](https://arxiv.org/abs/2510.26585)- SupervisorAgent is a lightweight, modular framework for runtime, adaptive supervision of multi-agent systems.

[LOOP Skill Engine](https://arxiv.org/abs/2605.14237)- LOOP records an agent's first run of a repetitive task with full LLM reasoning, then replays the extracted tool-call template without calling the LLM again.[Orchestrator-worker model tiering (frontier plans / cheap executes)](https://www.mindstudio.ai/blog/smart-orchestrator-cheaper-sub-agent-models-claude-code)- A capable model plans while cheaper agents execute; the pattern now ships as a vendor default, hitting 89.7% of LLM quality at 4% of the cost.[token-ninja](https://github.com/oanhduong/token-ninja)- Intercepts deterministic commands like git status or npm test before they reach the model, running them locally and skipping the LLM call.

[Claude Code v2.1.199 - transient-retry hardening & partial-output preservation](https://github.com/anthropics/claude-code/releases/tag/v2.1.199)- Claude Code v2.1.199 now auto-retries rate-limit errors with backoff and raised the default retry ceiling to 300, up from a prior cap of 15.

[Antigravity CLI - per-subagent model-tier routing + /effort (v1.1.5)](https://github.com/google-antigravity/antigravity-cli/blob/main/CHANGELOG.md)- Antigravity CLI v1.1.5 shipped first-party per-subagent model-tier routing (a model: flash|pro field in custom-agent frontmatter) plus an /effort control -…[Claude Code via a LiteLLM gateway (cheap-tier-in-front setup)](https://docs.litellm.ai/docs/tutorials/claude_non_anthropic_models)- Pointing Claude Code's ANTHROPIC_BASE_URL at a local LiteLLM proxy lets cheaper or non-Anthropic models absorb work the frontier model would otherwise bill for.[Cluster, Route, Escalate - cost-aware cascaded serving](https://arxiv.org/abs/2606.27457)- This paper proposes a two-stage cost-aware cascade for LLM serving that combines routing and escalation into one framework.[Distilling agent behavior into small task-specific models](https://arxiv.org/abs/2505.17612)- Distilling a large agent's behavior into a small 0.5-3B model lets most of its work run at a fraction of the frontier model's per-token cost.[MTRouter - per-turn cost-aware routing with history-model joint embeddings](https://arxiv.org/abs/2604.23530)- MTRouter picks a different model for each turn of a multi-turn conversation, rather than one model per query, to hit a cost budget without losing quality.[Not Diamond](https://www.notdiamond.ai/)- Not Diamond's meta-model predicts, per input, which LLM will give the best answer at the lowest cost, then routes the request there.[OpenCode - open-source coding-agent CLI with explicit cost-tier routing](https://opencode.ai/docs/)- OpenCode is an open-source (MIT) coding-agent CLI with its own explicit cost- and model-routing configuration, set directly in config.[OrcaRouter - production LinUCB bandit router (hybrid offline-online)](https://arxiv.org/abs/2605.30736)- OrcaRouter is a production LLM router built on a LinUCB bandit, with its cost/quality tradeoff independently confirmed on the RouterArena leaderboard.[RouteLLM](https://github.com/lm-sys/RouteLLM)- LMSYS's open-source router sending each query to a cheap or expensive model based on a trained cost threshold, as a drop-in server.[ruflo (formerly Claude-Flow; npm](https://github.com/ruvnet/ruflo)- ruflo is an open-source agent meta-harness for Claude Code and Codex, providing swarm orchestration and persistent memory.`claude-flow`

v3.17.0) - agent meta-harness with cost-adjusted model routing[vLLM Semantic Router](https://github.com/vllm-project/semantic-router)- Sends routine queries to cheap or local models and hard ones to stronger backends, as an open-source, self-hostable router.

[RLM-Cascade - response-level speculative decoding at the gateway](https://arxiv.org/abs/2606.22840)- RLM-Cascade, from a PayPal team, has a cheap draft model answer first and an Opus 4.8 verifier accept or rewrite it, at roughly 2% of Opus's cost.[SGLang](https://github.com/sgl-project/sglang)- A high-performance serving framework for large language and multimodal models.[vLLM](https://github.com/vllm-project/vllm)- The canonical open-source LLM serving engine, using PagedAttention to manage KV-cache memory in blocks so more requests batch at lower cost.

[Stop When Reasoning Converges](https://arxiv.org/abs/2605.17672)- Reasoning models often keep generating steps after a solution has already stabilized, wasting tokens and adding latency - what this paper calls "overthink.".[When more reasoning hurts - the test-time-compute ceiling](https://arxiv.org/abs/2604.10739)- Two 2026 papers found giving a model more reasoning budget makes it perform worse and cost more; past a point, tool delegation wins outright.

[Code execution with MCP (Anthropic)](https://www.anthropic.com/engineering/code-execution-with-mcp)- Anthropic proposes agents call MCP servers by writing and executing code instead of a tool call per step, so unused tool schemas skip the context window.[MCP Tool Descriptions Are Smelly!](https://arxiv.org/abs/2602.14878)- This study found poorly-written MCP tool descriptions measurably hurt agent efficiency, using an LLM-jury scanner and an A/B protocol on MCP-Universe.[Tool Attention Is All You Need](https://arxiv.org/abs/2604.21816)- MCP re-sends every tool's full schema on every turn, whether or not the agent needs it - a protocol tax known as the MCP/Tools Tax.

[CloudZero](https://www.cloudzero.com/blog/ai-cost-optimization-at-scale/)- An established commercial cloud and AI cost-intelligence / FinOps platform that brands itself 'The AI ROI Company'.[JetBrains AI moves business plans from monthly licenses to 12-month credits](https://blog.jetbrains.com/blog/2026/07/07/jetbrains-ai-for-teams-and-organizations-from-fragmented-ai-usage-to-coordinated-software-development/)- JetBrains is moving business AI from monthly per-seat licenses to 12-month reallocatable credits plus a governance dashboard; personal Pro/Ultimate prices are…[Mavvrik (fmr. DigitalEx)](https://www.mavvrik.ai/press-releases/mavvrik-unveils-full-stack-ai-cost-governance/)- Mavvrik is an AI/hybrid-infrastructure cost governance and FinOps platform, rebranded from DigitalEx in February 2025.[Pay-i](https://docs.pay-i.com/)- An SDK-based GenAI cost-observability platform that tracks token-level spend per call and rolls it up into cost-center allocation across orgs and apps.

[Denial-of-Wallet / token-exhaustion attacks](https://arxiv.org/abs/2601.10955)- Denial-of-wallet attacks exploit pay-per-token pricing to inflate a bill, via stolen-credential LLMjacking or agents steered into runaway token use.[Governance Decay - compaction silently erasing safety/governance constraints](https://arxiv.org/abs/2606.22528)- Compacting an agent's context can silently erase governance rules: across 7 model families, violations rose from 0% to 30%, up to 59% for some.

[FinOps for AI - canonical practitioner framework for governing AI/LLM spend](https://www.finops.org/framework/scope/finops-for-ai/)- FinOps for AI is the FinOps Foundation's official practitioner framework for governing AI, GPU, and token spend.[Vaudit - TokenAudit (LLM invoice reconciliation / AI spend audit)](https://www.vaudit.com/)- Vaudit is an AI-native, independent spend-auditing and recovery platform (San Francisco, founded late 2023).

[Claude Code's 5-hour/weekly usage quotas - Anthropic has stopped publishing exact numbers](https://support.claude.com/en/articles/11049741-what-is-the-max-plan)- Anthropic stopped publishing exact Claude Code usage quotas, describing Max plans only as 5x/20x multipliers of Pro with no absolute numbers.[TrueFoundry (AI Gateway - Budget Limiting)](https://www.truefoundry.com/docs/ai-gateway/budgetlimiting)- TrueFoundry is an enterprise GenAI deployment/gateway company founded by ex-Meta founders.

[CodeCarbon](https://github.com/mlco2/codecarbon)- An open-source (MIT) library for estimating a workload's energy use and CO2e emissions, and ML's widely-cited carbon baseline.[EcoLogits - energy/carbon of LLM](https://github.com/mlco2/ecologits)- Estimates the energy and carbon footprint of calling generative-AI APIs: the hosted counterpart to CodeCarbon, which measures your own hardware.*API*calls (the hosted-usage estimator)[Epoch AI - how much energy a query uses (the per-token energy anchor)](https://epoch.ai/gradient-updates/how-much-energy-does-chatgpt-use)- Epoch AI built a transparent, first-principles estimate of how much energy one LLM query costs.[Google - measuring the environmental impact of AI inference (provider disclosure)](https://cloud.google.com/blog/products/infrastructure/measuring-the-environmental-impact-of-ai-inference)- Google published a first-party disclosure of the energy, carbon, and water cost of a median Gemini Apps text prompt, authored by Amin Vahdat and Jeff Dean.[ML.ENERGY Leaderboard v3.0 - measured inference energy (the "reasoning ≈ 25× energy" signal)](https://ml.energy/blog/measurement/energy/diagnosing-inference-energy-consumption-with-the-mlenergy-leaderboard-v30/)- Version 3.0 of this leaderboard measures real GPU inference energy across 46 models x 7 tasks, finding reasoning models use roughly 25x the energy of others.

[ActPlane](https://github.com/eunomia-bpf/ActPlane)- An eBPF-based, OS-level policy-enforcement engine for AI-agent harnesses like Claude Code and Codex.[AEGIS](https://github.com/Justin0504/Aegis)- An open-source (MIT) pre-execution firewall and cryptographic audit layer for AI agents.[MCPGuard-Dynamic](https://github.com/facebook/mcpguard-dynamic)- An early-stage, research-grade kernel-level eBPF sandbox for MCP (64★), published under Meta's official GitHub org.

[ChatGPT Enterprise - usage analytics & spend controls](https://openai.com/index/chatgpt-enterprise-spend-controls/)- OpenAI's first-party spend layer for ChatGPT Enterprise/Business: a Global Admin Console with credit caps, request workflows, and a Cost API.[Claude Enterprise - admin analytics & cost controls](https://www.claude.com/blog/giving-admins-more-visibility-and-control-over-claude-usage-and-spend)- Anthropic's first-party spend surface for Claude Enterprise/Team admins: org-level spend caps, model defaults, and per-user cost analytics via the Admin API.[PointFive (AI Efficiency OS / TokenShift)](https://www.pointfive.co/press/pointfive-launches-ai-efficiency-os-tokenshift)- PointFive's TokenShift governs coding-agent token spend across Claude Code, Cursor, Codex, and more, claiming a 10-20% cut across 11 partners.[Revenium - runtime AI economic control / spend management](https://www.revenium.ai/)- Revenium tracks AI agent spend at runtime to the cent, attributing every model call and tool cost to its workflow, with auto-shutoff on runaway budgets.[Vantage](https://www.vantage.sh/blog/agentic-coding-costs)- A FinOps platform ingesting native token-level cost data from Anthropic and OpenAI's own usage APIs, plus Cursor and cloud spend.[Vercel AI Gateway - per-API-key budgets](https://vercel.com/changelog/budgets-for-api-keys-on-ai-gateway)- Vercel AI Gateway lets you cap spend per API key in dollars (min $1) with a daily/weekly/monthly refresh, rejecting further requests once the cap is hit.

[Claude "subscription arbitrage" and its (announced, then paused) end](https://zed.dev/blog/anthropic-subscription-changes)- Users route agentic workloads worth far more than a subscription's price through cheap Pro/Max plans; Anthropic tried to close this, then paused the fix.[Coding-agent native spend controls (2026)](https://cursor.com/changelog/05-04-26)- Within six weeks in 2026, Cursor, GitHub Copilot, and OpenAI each shipped native admin spend controls: budget caps, credit metering, usage dashboards. (also:[GitHub Copilot](https://github.blog/news-insights/company-news/github-copilot-is-moving-to-usage-based-billing/)·[OpenAI](https://openai.com/index/chatgpt-enterprise-spend-controls/))[GitHub Copilot metered-billing bill-shock - the demand-side reaction ("tokenpocalypse")](https://news.ycombinator.com/item?id=47923357)- GitHub's move from flat-rate Copilot plans to metered AI Credits exposed agentic workflows' true per-token cost and triggered a mass bill-shock backlash.[Lanai](https://www.prnewswire.com/news-releases/lanai-launches-ai--work-operating-system-to-help-enterprises-close-the-ai-accountability-gap-302743892.html)- Lanai's AI @ Work platform discovers every sanctioned and shadow AI workflow across an org and maps its token spend to the KPIs it actually drives.[State of FinOps 2026 - AI spend management is now the norm](https://data.finops.org/)- This is the FinOps Foundation's sixth annual State of FinOps survey, the practitioner census of how organizations manage cloud and AI spend.[The "$47k Claude Code bill" - the anchor bill-shock anecdote and its mechanistic debunk](https://yusufhansacak.medium.com/the-47-000-agent-bill-what-the-viral-token-stories-get-wrong-7ee1cdd81e65)- A viral $47,000-in-90-days Claude Code bill story was debunked by a teardown pinning the real driver on quadratic context re-ingestion, not runaway use.[Uber caps AI-coding spend at $1,500/mo per tool after burning its budget in ~4 months](https://techcrunch.com/2026/06/02/uber-caps-employee-ai-spending-after-blowing-through-budget-in-four-months/)- Uber capped AI-coding spend at $1,500 per employee per tool after burning its entire annual budget in roughly four months of encouraged maximal use.

[JetBrains independently measured two token-saving skills against their own claims](https://blog.jetbrains.com/ai/2026/07/rtk-claude-code-token-savings/)- JetBrains independently A/B-tested two token-saving skills: rtk ran +7.6% more expensive at low effort (claimed 60-90% cut), Caveman saved ~8.5% (claimed 65%). (also:[Caveman A/B post](https://blog.jetbrains.com/ai/2026/07/speak-to-ai-agents-like-cavemen-tosave-tokens/))[Token-saving plugins are mostly a stupid idea (Tura benchmark)](https://turaai.net/blog#token-saving-plugins-are-mostly-stupid-idea)- A benchmark of token-saving plugins found one actively worse than none: cost up 7.2%, tokens up 13.2%, because it broke an already-cached prompt prefix.

[Cisco acquires Galileo (LLM eval/observability) → folded into Splunk Observability](https://blogs.cisco.com/news/cisco-announces-the-intent-to-acquire-galileo)- Cisco acquired Galileo, an LLM/agent evaluation and observability platform, folding it into Splunk Observability Cloud's AI Agent Monitoring.[Tokenomics Foundation (Linux Foundation + FinOps Foundation)](https://www.linuxfoundation.org/press/linux-foundation-announces-the-intent-to-launch-the-tokenomics-foundation-to-establish-open-standards-for-ai-cost-management)- The Linux Foundation is launching the Tokenomics Foundation to build open standards for AI token spend, extending FOCUS to cover token-based costs.

[Aider - an OSS coding CLI that meters its own dollar cost](https://github.com/Aider-AI/aider)- An open-source terminal coding agent (Apache-2.0, ~48k★) with built-in per-message dollar-cost tracking and a public polyglot leaderboard that publishes…[Amp (Sourcegraph) - pay-as-you-go, no-markup pricing + mode-based routing](https://ampcode.com/)- Amp, Sourcegraph's coding agent, passes through LLM cost with zero markup for individuals and teams, with a cost/capability mode: Deep, Smart, or Rush.[Cline - an OSS coding agent on a bring-your-own-key cost model](https://github.com/cline/cline)- An open-source AI coding agent (Apache-2.0, ~65k★, ~4.8M VS Code installs), built on a bring-your-own-API-key cost model with per-task token and dollar…[Factory (droids) - subscription pricing + $150M Series C at $1.5B](https://factory.ai/pricing)- Factory prices its Droids agents as flat subscription tiers ($20/$100/$200/mo) with usage-based rate limits, not per-token metering, after a $150M round.[Gemini CLI retirement → Antigravity CLI (open-source coding agent closes, pricing restructures)](https://developers.googleblog.com/an-important-update-transitioning-gemini-cli-to-antigravity-cli/)- Google retired the open-source Gemini CLI on 2026-06-18, pushing users onto closed-source Antigravity CLI and $100/$200-per-month paid tiers.[OpenHands - MIT OSS coding agent, free local + free cloud tier, at-cost LLM option](https://github.com/OpenHands/OpenHands)- An MIT-licensed open-source coding-agent platform with a free local mode, a free cloud tier, and an at-cost LLM pricing option.

[Gartner - worldwide AI spending forecast: $2.59T in 2026 (+47% YoY)](https://www.gartner.com/en/newsroom/press-releases/2026-05-19-gartner-forecasts-worldwide-ai-spending-to-grow-47-percent-in-2026)- Gartner's latest forecast puts worldwide AI spending at $2.59 trillion in 2026, up 47% year-over-year, with infrastructure over 45% of the total.[Menlo Ventures - enterprise generative-AI spend $11.5B → $37B (2024→2025)](https://menlovc.com/perspective/2025-the-state-of-generative-ai-in-the-enterprise/)- Menlo Ventures found enterprise generative-AI spend hit $37B in 2025, up 3.2x from 2024, with coding tools the largest application category at $7.3B.

["Qwen 3.6 27B is the sweet spot for local development" - Migdał / Quesma (first-party)](https://quesma.com/blog/qwen-36-is-awesome/)- Piotr Migdał's #1-on-Hacker-News essay argues Qwen3.6-27B (dense) is the first local model good enough for real coding instead of a metered cloud API.[Artificial Analysis - Coding Agent Index (tokens & cost per task, model × harness)](https://artificialanalysis.ai/agents/coding-agents)- This benchmark scores full model-plus-harness stacks, spanning $0.27 to $11.80 per task: a roughly 44x range at similar quality, per Artificial Analysis.[Artificial Analysis - Intelligence Index + Blended Price (intelligence-per-dollar leaderboard)](https://artificialanalysis.ai/leaderboards/models)- Artificial Analysis's Intelligence Index plots a 0-100 capability score against blended price per million tokens, live across 85-122 base LLMs.[Claude Opus 5 - flat price vs Opus 4.8, but 1M context and thinking on by default](https://platform.claude.com/docs/en/release-notes/api)- Claude Opus 5 launched 2026-07-24 at the same $5/$25 per MTok as Opus 4.8, but ships 1M context and thinking on by default - so the flat per-token price hides a…[Gemini 3.6 Flash - a Flash tier marketed on fewer tokens per task, not just a lower unit price](https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-6-flash-3-5-flash-lite-3-5-flash-cyber/)- Gemini 3.6 Flash (2026-07-21) is priced lower than 3.5 Flash at $1.50/$7.50 per MTok AND uses 17% fewer output tokens on the same work - a rare launch where the…[Kimi K2.6/K2.7-Code and GLM-5.2 official API pricing](https://platform.kimi.ai/docs/pricing/chat-k27-code)- Kimi K2.7-Code ($0.95/$4.00 per million tokens) and GLM-5.2 ($1.40/$4.40) both undercut Claude Sonnet 5 and GPT-5.5 on raw price by 2-5x.[Local / open-model economics for coding - state of the field (2026)](https://huggingface.co/Qwen/Qwen3.6-27B)- Open-weight coding models now score 77-81% on SWE-bench Verified, within a few points of closed frontier models, reshaping self-host-vs-API math.[OckBench - measuring token efficiency / verbosity of LLM reasoning](https://arxiv.org/abs/2511.05722)- OckBench answers a specific tokenomics question: which model burns the most tokens for the same answer?[Reasoning-token billing across providers - the hidden output multiplier](https://developers.openai.com/api/docs/guides/reasoning)- Every major AI provider bills a model's hidden reasoning tokens at the most expensive output rate, without ever returning them to the caller. (also:[Google](https://ai.google.dev/gemini-api/docs/pricing)·[DeepSeek](https://api-docs.deepseek.com/guides/reasoning_model))[Reasoning-token consumption behavior - length ≠ effort, and verbosity is a separate lever](https://arxiv.org/abs/2602.13517)- Chain-of-thought can burn about 258 tokens on problems a direct answer solves in 15 (roughly 17x overhead), and simple agentic steps trigger it by accident.

[Batch / Priority / Flex service tiers - the scheduling axis of token pricing (clustered, cross-vendor)](https://platform.claude.com/docs/en/docs/build-with-claude/batch-processing)- Every major LLM vendor sells the same lever, trading latency for price via async batch scheduling, with Anthropic, OpenAI, and Google all near 50% off. (also:[OpenAI](https://developers.openai.com/api/docs/pricing)·[Google](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/claude/batch))[Bessemer - the AI pricing & monetization playbook (seat → usage → outcome)](https://www.bvp.com/atlas/the-ai-pricing-and-monetization-playbook)- Bessemer's playbook argues AI pricing is shifting from per-seat to consumption/outcome-based, citing Intercom's $0.99-per-resolved-ticket model.[Cached-input discounts - the ~90%-off lever behind cache-accounting](https://platform.claude.com/docs/en/build-with-claude/prompt-caching)- Cache-read pricing discounts input tokens by about 90%: the biggest lever on an agentic bill, since input is roughly 85% of session cost.[ChatGPT subscription tiers and Codex CLI bundling/pricing (2026)](https://developers.openai.com/codex/pricing)- OpenAI bundles Codex CLI into every ChatGPT tier from Free through the new $100/mo Pro plan, differing only by rate-limit multiplier.[ChatGPT workspace-agent credit billing (effective July 6, 2026)](https://help.openai.com/en/articles/11481834-chatgpt-rate-card)- OpenAI ended the free preview for agent runs invoked inside ChatGPT Business, Enterprise, Edu, and Teachers on 2026-07-06.[Cursor charges by tokens, split into first-party and third-party pools](https://cursor.com/docs/account/pricing)- Cursor meters by tokens per million (input/output/cache-write/cache-read), split into a first-party pool and a third-party API pool, with a $0.25/M Teams… (also:[Teams pricing blog](https://cursor.com/blog/teams-pricing-june-2026))[Devin's Agent Compute Unit has no published definition of what it meters](https://docs.devin.ai/admin/billing/enterprise)- Devin bills Enterprise usage in Agent Compute Units, but no official doc defines what an ACU measures (not tokens, seconds, or calls); the opacity itself is the…[Fable 5 leaves subscription inclusion - frontier tier moves to usage-credit metering (July 7 cliff)](https://www.anthropic.com/news/redeploying-fable-5)- Fable 5's subscription saga settled 2026-07-20 (after two extensions) as a primary-confirmed two-tier split: Max & Team Premium keep it included (50% of weekly…[Google AI Pro price and Gemini/Antigravity free-tier limits (2026)](https://gemini.google/subscriptions/)- Google AI Pro is confirmed at $19.99/month, beneath the $99.99 and $199.99 AI Ultra tiers giving higher rate limits on the Gemini API and Antigravity.[GPT-5.6 family (Sol / Terra / Luna) - API pricing](https://developers.openai.com/api/docs/pricing)- OpenAI's GPT-5.6 family prices three tiers 2x apart: Sol at $5/$30 per million tokens, Terra at $2.50/$15, and Luna at $1/$6.[LiteLLM flex/priority service-tier cost keys - the harness-level tier-routing lever](https://docs.litellm.ai/docs/proxy/custom_pricing)- LiteLLM automatically prices requests made at a non-standard tier like flex or priority, applying the right discounted or premium rate automatically.[LLM price decline + Jevons paradox - unit price crashes, total spend climbs](https://a16z.com/llmflation-llm-inference-cost/)- Per-token prices are falling roughly an order of magnitude per year, while total AI spend rises even faster.[LLM token pricing dimensions - the structure of a token bill](https://platform.claude.com/docs/en/about-claude/pricing)- This maps out how frontier LLM APIs meter and price tokens, read straight off the two largest providers' pricing pages, Anthropic and OpenAI.[OpenAI is winding down the self-serve fine-tuning API and platform](https://developers.openai.com/api/docs/deprecations)- OpenAI is winding down self-serve fine-tuning because prompting got cheaper and more capable than fine-tuning for most uses, cutting off customers by 2027.[Tokenization multiplicity & overcharging - the pay-per-token integrity problem](https://arxiv.org/abs/2506.06446)- Two academic papers show the same output can be billed a different token count depending on tokenization, and providers can be incentivized to inflate it.[Windsurf became Devin Desktop and switched credits to token-based quota](https://docs.devin.ai/desktop/accounts/quota)- Windsurf became Devin Desktop and in March 2026 swapped opaque per-model credit multipliers for token-based quota where free models cost nothing; a…

[Reserved-capacity reliability economics (Azure PTU · AWS Bedrock MU)](https://learn.microsoft.com/en-us/azure/ai-foundry/openai/concepts/provisioned-throughput)- Azure's Provisioned Throughput Units and AWS Bedrock's Model Units both let buyers reserve guaranteed capacity, billed hourly whether or not it's used.

[Cloud Capital - gross margin in the age of AI (the vendor/supply side)](https://www.cloudcapital.co/learn/gross-margin-in-the-age-of-ai)- AI-native software runs at roughly 50-60% gross margin versus 70-80% for SaaS, since inference and compute became a large, variable cost of goods sold.[Cost-of-Pass - an economic framework for evaluating language models](https://arxiv.org/abs/2504.13359)- Cost-of-Pass defines the expected dollar cost of one correct answer as inference cost divided by success rate, pricing benchmark accuracy directly.[DORA 2025 - AI as amplifier, and the delivery-stability tension](https://dora.dev/insights/balancing-ai-tensions/)- Google's DORA program found that as AI adoption becomes universal, delivery throughput rises but so does instability: AI as an amplifier, not a pure win.[Faros - "The Acceleration Whiplash" (AI Engineering Report 2026)](https://pages.faros.ai/hubfs/AI_Engineering_Report_2026_The_Acceleration_Whiplash_Faros.pdf)- The "velocity has a hidden bill" study: telemetry from 22,000 developers across 4,000 teams over two years.[getDX - AI coding assistant pricing & ROI guide (2026)](https://getdx.com/blog/ai-coding-assistant-pricing/)- Typical AI coding tools cost $200-600 per engineer monthly in seat plus token spend, per getDX, for a median 7.76% PR gain: below vendors' claimed 3-10x.[METR - measured vs perceived AI productivity (the RCT)](https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/)- METR's controlled trial found developers took 19% longer to finish issues when allowed to use AI, while still believing it had sped them up by 20%.

[Claw-SWE-Bench](https://github.com/opensquilla/claw-swe-bench)- Found that adapter/harness design alone swings an agent's Pass@1 score by about 54 percentage points on the identical model backbone.[Coding Benchmarks Are Misaligned with Agentic SE (Tessl)](https://arxiv.org/abs/2606.17799)- A position paper from Tessl (London, UK) argues that today's coding benchmarks don't measure what people think they measure.[Deterministic Anchoring - how much static structure do code agents need?](https://arxiv.org/abs/2606.26979)- This ISSTA 2026 paper found injecting static-analysis facts as plain-text comments raises a code agent's Pass@1 by 3.4pp and cuts trajectories by 1.6 rounds.[GitHub Copilot agentic-harness efficiency evaluation (first-party offline ablation)](https://github.blog/ai-and-ml/github-copilot/evaluating-performance-and-efficiency-of-the-github-copilot-agentic-harness-across-models-and-tasks/)- GitHub's own benchmark plots Copilot's agentic-harness resolution rate against dollar-cost-per-task across five benchmarks and four frontier models.[Harness-Bench](https://arxiv.org/abs/2605.27922)- Holds the task, model, and budget fixed while varying only the agent harness, across 5,194 trajectories spanning 6 harnesses and 8 models.[LoCoMo](https://github.com/snap-research/locomo)- Snap Research's very-long-term conversational-memory benchmark; the canonical dataset that Mem0, Zep, and Supermemory all cite, kept as a background instrument…[LongMemEval](https://github.com/xiaowu0162/LongMemEval)- The peer-reviewed (ICLR 2025) benchmark for long-term memory in chat assistants; the standard eval that mem0, Zep, and Supermemory all self-report their…[MemoryBench](https://github.com/supermemoryai/memorybench)- A pluggable harness to run memory systems (Supermemory, Mem0, Zep) head-to-head across datasets like LoCoMo; useful for standardizing comparison, but built by…[Prompt Compression in the Wild - the end-to-end referee for compression](https://arxiv.org/abs/2604.02985)- This ECIR 2026 study found LLMLingua's compression yields up to 18% speed-up only in a narrow window; outside it, the compression step cancels the gains.[promptfoo](https://github.com/promptfoo/promptfoo)- An open-source CLI/CI harness for testing LLM prompts and agents that records per-eval token usage and cost as an assertable metric.[RedundancyBench - can anyone even](https://arxiv.org/abs/2605.29893)- RedundancyBench is a benchmark for step-level redundancy detection in agent trajectories - can a model even spot the wasted step in an agent's history?*detect*a redundant step?[RouterArena - the open benchmark + live leaderboard for LLM routers](https://github.com/RouteWorks/RouterArena)- RouterArena is an open evaluation platform and live leaderboard for LLM routers - systems that auto-select a model per query.[SWE-bench - the root benchmark the cost-aware derivatives build on](https://github.com/SWE-bench/SWE-bench)- The canonical accuracy-only software-engineering benchmark (2,294 real GitHub issue tasks, 12 Python repos, ICLR 2024) that every cost-aware derivative in this…[SWE-Effi - cost-aware re-ranking of SWE-agents under resource budgets](https://arxiv.org/abs/2509.09853)- SWE-Effi re-ranks popular AI issue-resolution systems on a SWE-bench subset by cost-under-resource-constraints instead of by accuracy alone.[Terminal-Bench](https://www.tbench.ai/)- The canonical benchmark for AI agents in real terminal/CLI environments, with 89 tasks each vetted through ~3 reviewer-hours.

[Gemini context caching - the per-hour storage meter (the third-vendor axis)](https://ai.google.dev/gemini-api/docs/pricing)- Among major providers, only Gemini bills a per-hour storage meter for explicit prompt caching, making a cache a rented line item that accrues cost while it sits…[OpenAI prompt-caching](https://platform.openai.com/docs/guides/prompt-caching)- This is the measurement substrate for prompt-cache savings on the OpenAI API.`cached_tokens`

accounting[OpenAI Responses conversation state - you pay for the whole chain every turn (and the compaction levers)](https://developers.openai.com/api/docs/guides/conversation-state)- This entry covers the billing semantics of stateful conversations on OpenAI's Responses API, plus the two server-side compaction levers that mitigate them.

[Anthropic bill anatomy - the whole-bill line-item taxonomy](https://platform.claude.com/docs/en/about-claude/pricing#code-execution-tool)- This is the canonical enumeration of every meter on a Claude API bill, read live from Anthropic's pricing page.[TensorZero - cross-vendor token-count divergence ("stop comparing $/M tokens")](https://www.tensorzero.com/blog/stop-comparing-price-per-million-tokens-the-hidden-llm-api-costs/)- The same input can produce 2.65x more tokens on one tokenizer than another's: Claude Opus 4-7 runs 1.57x-2.65x more tokens than GPT-5.4 on the same content.[Vision-token pricing formulas across the big three](https://platform.claude.com/docs/en/build-with-claude/vision)- Anthropic, OpenAI, and Google each convert an image into billed tokens with a different formula, so no single cross-vendor image-cost number exists. (also:[OpenAI](https://developers.openai.com/api/docs/guides/images-vision)·[Google](https://ai.google.dev/gemini-api/docs/image-understanding))

[Claude Code system prompts (Piebald extraction)](https://github.com/Piebald-AI/claude-code-system-prompts)- Piebald AI extracts Claude Code's full compiled prompt payload per release: 515 prompt strings and 27 tool descriptions at v2.1.212, each priced in tokens.[Claude Code vs OpenCode token overhead (Systima study)](https://systima.ai/blog/claude-code-vs-opencode-token-overhead)- Systima measured harness scaffolding overhead before a prompt is even read: Claude Code carries about 32,800 tokens versus OpenCode's 6,900, a 4.7x gap.

[Cross-vendor coding-agent usage trackers (AgentsView · caut)](https://github.com/kenn-io/agentsview)- AgentsView and caut are open-source tools that read local session logs to aggregate token usage and cost across roughly 20 coding-agent vendors.[How Do AI Agents Spend Your Money?](https://arxiv.org/abs/2604.22750)- This Stanford study is the first systematic look at token spend in agentic coding, running 8 frontier models on 500 SWE-bench Verified tasks.[tokview](https://github.com/headroomlabs-ai/tokview)- A local, zero-config proxy showing a coding agent's token spend by session, model, and tool call, flagging re-sent results that multiply the bill.

[Early Diagnosis of Wasted Computation in Multi-Agent LLM Systems (Failure-Aware Observability)](https://arxiv.org/abs/2606.01365)- This live waste-detection framework for multi-agent systems found that 58.1% of tokens in failed runs are spent after its first warning fires.[Faros AI - "Token Intelligence" / Token Attribution Ledger](https://www.faros.ai/blog/token-intelligence-for-ai-engineering)- Faros AI is an enterprise engineering-intelligence SaaS (DORA/SPACE-style productivity analytics).[token-optimizer (alexgreensh) - per-session $ waste-detection plugin, semi-cross-vendor](https://github.com/alexgreensh/token-optimizer)- token-optimizer is a coding-agent plugin that runs eleven heuristic waste detectors per session and prices the flagged tokens in dollars.

[FOCUS 1.4 - the cross-vendor billing-data normalization standard (now with invoice reconciliation)](https://focus.finops.org/focus-specification/)- FOCUS 1.4, the Linux Foundation's billing-data schema, added Invoice Detail and Billing Period datasets to reconcile spend against real invoices.

Tool-agnostic, evidence-grounded standards for token-efficient agentic coding. Each is one page: TL;DR, claim, evidence, links. [Browse the practices](/QuesmaOrg/awesome-ai-tokenomics/blob/main/practices/README.md).

Short reference notes explaining the ideas behind the practices: cache economics, the harness-waste taxonomy, orchestration economics. [Browse the concepts](/QuesmaOrg/awesome-ai-tokenomics/blob/main/concepts/README.md).

Confidence-scored beliefs, clearly labeled as beliefs rather than facts, each with its strongest evidence linked. [Read the claims](/QuesmaOrg/awesome-ai-tokenomics/blob/main/claims.md).

Runnable, validated Claude Code and Codex configurations and skills for token-efficient agentic coding, each labeled with how it was validated. [Browse the setups](/QuesmaOrg/awesome-ai-tokenomics/blob/main/setups/README.md).

Text content: CC-BY 4.0 ([LICENSE](/QuesmaOrg/awesome-ai-tokenomics/blob/main/LICENSE)) · Code and configs: MIT ([LICENSE-CODE](/QuesmaOrg/awesome-ai-tokenomics/blob/main/LICENSE-CODE)). Maintained by the team at [Quesma](https://quesma.com).