{"slug": "same-deepseek-v4-flash-different-agent-why-the-runtime-changes-the-result", "title": "Same DeepSeek V4 Flash, Different Agent: Why the Runtime Changes the Result", "summary": "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 effective unit is not the model ID but the complete runtime—model, protocol, tools, context, recovery, and acceptance—and proposes a formula: Effective Agent = Model potential × Harness realization rate. The developer calls for controlled comparisons that fix the backend model and other variables, warning that otherwise results are runtime observations, not model leaderboards.", "body_md": "Same DeepSeek V4 Flash. Different runtime. Very different long-task outcomes.\n\nMy local sample is bounded: Codex + Flash completed a long, cross-file, repeatedly verified deck task; Claude Code + Flash launched multiple reviews, but their quality was not independently verified. This supports different pairings, not a universal ranking.\n\nThe useful unit is not a model ID. It is a complete runtime: **model × protocol × tools × context × recovery × acceptance**.\n\n**Protocol is a trajectory interface.** It defines how goals, tool results, intermediate state, and continuation are represented. A compatibility layer can connect successfully and still lose long-horizon affordances.\n\n**Tools are contracts, not buttons.** Schemas, parameters, return formats, and failure signals define the action space. The same “read” or “edit” label can behave differently across runtimes.\n\n**Context and recovery make local intelligence durable.** A long task needs to preserve constraints, retain failure evidence, detect drift, return to a stable point, and re-plan.\n\n**Acceptance defines done.** An agent’s “done” is a self-report. Delivery means that file state, tests, previews, permissions, and external facts agree with the original goal.\n\nDeepSeek’s public update names its Harness minimal mode for code-agent benchmarks, says V4 Flash was adapted for Codex, and documents Responses API support. Public Code Harness hiring language reinforces the strategic direction. These are product signals, not proof of universal superiority.\n\nI write the idea as:\n\nEffective Agent = Model potential × Harness realization rate\n\nThe realization rate can be decomposed into protocol matching, tool-contract reliability, context/recovery quality, and acceptance evidence. A more credible comparison fixes the backend model ID, task, code state, client version, effort, permissions, and acceptance criteria; repeats the task; then records tool errors, rework, human intervention, and rollback.\n\nIf those conditions cannot be fixed, call the result a runtime observation—not a model leaderboard.\n\nThe model may set the ceiling. The runtime determines how much of that ceiling reaches the work.\n\nWhat is the thinnest layer in the agent you use today?\n\nDisclosure: This article was written with AI assistance and reviewed by the author.", "url": "https://wpnews.pro/news/same-deepseek-v4-flash-different-agent-why-the-runtime-changes-the-result", "canonical_source": "https://dev.to/octoooo/same-deepseek-v4-flash-different-agent-why-the-runtime-changes-the-result-3m3b", "published_at": "2026-08-01 15:11:59+00:00", "updated_at": "2026-08-01 15:50:36.789533+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-agents", "ai-research"], "entities": ["DeepSeek", "V4 Flash", "Codex", "Claude Code", "Responses API", "Code Harness"], "alternates": {"html": "https://wpnews.pro/news/same-deepseek-v4-flash-different-agent-why-the-runtime-changes-the-result", "markdown": "https://wpnews.pro/news/same-deepseek-v4-flash-different-agent-why-the-runtime-changes-the-result.md", "text": "https://wpnews.pro/news/same-deepseek-v4-flash-different-agent-why-the-runtime-changes-the-result.txt", "jsonld": "https://wpnews.pro/news/same-deepseek-v4-flash-different-agent-why-the-runtime-changes-the-result.jsonld"}}