OpenAI has deprecated its Assistants API, scheduled to shut down on August 26, 2026, and directs developers to migrate to the Responses API and Conversations API. The migration requires more than swap…
A new paper and online discussions have revealed methods to extract hidden reasoning traces from closed-weight AI models, prompting an investigation into how these traces work. Reasoning traces are si…
A developer at Benchclaw.io reports that progressive disclosure of tool schemas in an AI agent cut input tokens by about 30% and total cost by about 21%, but a task-level breakdown revealed that the a…
DeepSeek has released DeepSeek-V4-Pro-0813, an advanced AI model with enhanced agent capabilities, now available via its website, mobile app, and API. The model scored 62.7 on the DeepSWE benchmark, u…
DeepSeek has released its V4 Pro model with 1.6 trillion total parameters and 49 billion active per token, now generally available under an MIT license across its web platform, mobile app, and API. Th…
OpenAI will retire its Assistants API on 26 August 2026, a year after the announcement, forcing developers to migrate to the Responses API. A developer's analysis of npm download data shows that 99% o…
OpenAI will shut down its Assistants API on August 26, 2026, affecting all endpoints including /v1/assistants, /v1/threads, and /v1/threads/runs, with no grace period or extension. The company will no…
OpenAI's Responses API offers developers a task-oriented alternative to Chat Completions, centering workflows on response objects with structured outputs, tool events, and traceable execution rather t…
Amazon Web Services has expanded the context window for OpenAI's GPT-5.6 Sol, Terra, and Luna models on its Bedrock platform to 1 million tokens, effective immediately, enabling enterprises to process…
A developer's analysis shows that the same DeepSeek V4 Flash model produces different long-task outcomes depending on the agent runtime, such as Codex versus Claude Code. The developer argues that the…
DeepSeek has launched the public beta of its V4-Flash API, featuring upgrades for agent tasks, scoring 82.7 on Terminal Bench 2.1 and 54.4 on DeepSWE. The update adds Responses API support and Codex a…
OpenAI claims its GPT-5.6 Sol model scored 38.3% on the ARC-AGI-3 public set with two API settings enabled—retained reasoning and compaction—tripling its baseline score and surpassing Anthropic's Clau…
OpenAI reported that GPT-5.6 Sol's score on the ARC-AGI-3 benchmark nearly tripled from 13.3% to 38.3% after replacing the default test harness with two settings from its Responses API: retained reaso…
OpenAI core products lead Thibault Sottiaux said GPT-5.6 Sol reached a state-of-the-art score on ARC-AGI-3 after the company changed how the model's reasoning and context were carried between actions,…
A developer compares the OpenAI and Anthropic APIs for production SaaS features, arguing that API design—not model benchmarks—determines long-term maintainability and cost. The analysis highlights dif…
OpenAI shipped GPT-5.6 on July 9 as a tiered model family with three variants — Sol, Terra, and Luna — priced at $5/$30, $2.50/$15, and $1/$6 per million input/output tokens respectively, all sharing …
OpenAI introduces compaction for its Responses API, enabling developers to reduce context size in long-running interactions while preserving key state and reasoning. Server-side compaction can be enab…
OpenAI shipped GPT-5.6 on July 9 with Programmatic Tool Calling, a feature that lets the model write JavaScript to orchestrate tool calls in a V8 sandbox, reducing tokens by up to 63.5% and model turn…
OpenAI's GPT-5.6, generally available since July 9, 2026, introduces Programmatic Tool Calling that lets the model write JavaScript to orchestrate tools inside an isolated V8 sandbox, cutting round tr…
OpenAI shipped three permanent model tiers on July 9 — Sol, Terra, and Luna — under the GPT-5.6 generation, replacing the previous version-number churn with durable capability tiers that evolve indepe…