{"collection":{"slug":"frontier-ai-labs","name":"Frontier AI Labs"},"digest":{"id":"cwd_y1BcxLH9NSecyUQcaFX3h","weekStart":"2026-06-08","title":"Fable 5 debuts, then vanishes — and tool transparency deepens","intro":"Anthropic launched Claude Fable 5 and Mythos 5 with 1M token context, then had access suspended by a US government export directive by week's end. Across both Anthropic and OpenAI, tooling gained sharper controls: Claude Code sub-agents can now spawn sub-agents, web search tools expose response-inclusion parameters, and ChatGPT added rate-limit reset banking.","body":"### A model week turned upside down\n\nMonday brought what looked like the biggest launch of the quarter. Anthropic released [Claude Fable 5 and Mythos 5](/release/rel_Afd7WU3ikzXvrFx31z9wB-claude-platform-launches-claude-fable-5-and-claude-mythos-5-with-1m-token) — both with a 1M-token context window by default, 128k max output tokens, and always-on adaptive thinking that can no longer be disabled. A new tokenizer produces roughly 30% more tokens for the same text, and safety classifiers now run on every request and during response generation, with an opt-in fallback that re-runs refused requests on another model. By Friday, the picture had flipped entirely. The US government issued [an export-control directive suspending all access to Fable 5 and Mythos 5](/release/rel_wFBOY7nTpVVtkEMAaTlQn-statement-on-the-us-government-directive-to-suspend-access-to-fable-5-and-mythos), and Anthropic confirmed access was [effectively suspended](/release/rel_CPkcfiokKwF8tMQshREnn-claude-and-claude-mythos-5-access-suspended). Meanwhile, older model versions were retired: Claude Sonnet 4 and Opus 4 were [taken offline](/release/rel_t6-BYihayp9tRzsaABpHs-claude-sonnet-4-and-claude-opus-4-models-retired), and OpenAI also phased out [GPT-5.2 models in ChatGPT](/release/rel_RGLYUx7d7Li4vsN7uyQ4q-chatgpt-retires-gpt-5-2-models), auto-migrating conversations to GPT-5.5.\n\n### Tool transparency and tighter agent controls\n\nWhile model access shifted abruptly, the platform tools powering agentic workflows saw quieter but important polish. The [code execution tool](/release/rel_ovBDJWt_FStOhpw_biWj8-claude-platform-code-execution-tool-now-discloses-90-second-cell-time-limit) now explicitly advertises its 90-second per-cell time limit in the tool description, giving Claude the information it needs to budget long-running cells. The [web search and web fetch tools](/release/rel_HKLjAFmKc6Mvhsqe8UHX--claude-platform-adds-response-inclusion-parameter-to-web-search-and-fetch-tools) gained a `response_inclusion` parameter that lets agentic workflows drop consumed result blocks from the API response, reducing wire noise for multi-step search loops.\n\nFor self-hosted sandbox builders on AWS, a new [`GET /v1/environments/{id}/work` endpoint](/release/rel_DXYaPjyG4su2fGyysx5FN-claude-platform-on-aws-adds-get-v1-environments-id-work-endpoint) lets infrastructure poll pending work items, authorized through the `GetEnvironment` IAM action.\n\nClaude Code v2.1.172 crossed a practical threshold: [sub-agents can now spawn their own sub-agents](/release/rel_pOYmtmJpaTpXCZwi7Y5Z0-claude-code-v2-1-172-allows-sub-agents-to-spawn-sub-agents-and-fixes-context), up to five levels deep. Subsequent patches fixed Fable 5 model name normalization and added an `enforceAvailableModels` managed setting that constrains the default model to the admin allowlist.\n\n### OpenAI ecosystem: partner network and user-visible controls\n\nOpenAI formalized its go-to-market structure by launching the [OpenAI Partner Network](/release/rel_hW5VT0pKzOL2qDpK0Y1O1-openai-launches-partner-network-to-scale-ai-adoption), a program backed by $150M to help partners build, sell, and deploy AI solutions. On the product side, ChatGPT added [rate-limit reset banking](/release/rel_HOLmi6zZTBBzrOqy5C4ig-codex-2026-06-11-app-adds-rate-limit-reset-banking-and-developer-mode) for Plus and Pro users, and a Developer mode for the in-app browser that exposes controlled Chrome DevTools Protocol access for debugging. The [ChatGPT memory management page](/release/rel_R3D1VN_awz-thGK7CgeO7-chatgpt-adds-controls-to-delete-memories-and-turn-memory-off) now lets users delete individual memories or turn memory off entirely.","releaseIds":["rel_Afd7WU3ikzXvrFx31z9wB","rel_wFBOY7nTpVVtkEMAaTlQn","rel_CPkcfiokKwF8tMQshREnn","rel_t6-BYihayp9tRzsaABpHs","rel_RGLYUx7d7Li4vsN7uyQ4q","rel_ovBDJWt_FStOhpw_biWj8","rel_HKLjAFmKc6Mvhsqe8UHX-","rel_DXYaPjyG4su2fGyysx5FN","rel_pOYmtmJpaTpXCZwi7Y5Z0","rel_hW5VT0pKzOL2qDpK0Y1O1","rel_HOLmi6zZTBBzrOqy5C4ig","rel_R3D1VN_awz-thGK7CgeO7"],"releaseCount":55,"generatedAt":"2026-07-11T16:38:51.571Z","releases":[{"id":"rel_Afd7WU3ikzXvrFx31z9wB","title":"Claude Platform launches Claude Fable 5 and Claude Mythos 5 with 1M token context","path":"/release/rel_Afd7WU3ikzXvrFx31z9wB-claude-fable-5-and-mythos-5-ship-1m-context-and-adaptive-thinking","org":{"slug":"anthropic","name":"Anthropic"},"importance":5},{"id":"rel_wFBOY7nTpVVtkEMAaTlQn","title":"Statement on the US government directive to suspend access to Fable 5 and Mythos 5","path":"/release/rel_wFBOY7nTpVVtkEMAaTlQn-statement-on-the-us-government-directive-to-suspend-access-to-fable-5-and-mythos","org":{"slug":"anthropic","name":"Anthropic"},"importance":null},{"id":"rel_CPkcfiokKwF8tMQshREnn","title":"Claude and Claude Mythos 5 access suspended","path":"/release/rel_CPkcfiokKwF8tMQshREnn-claude-and-claude-mythos-5-access-suspended","org":{"slug":"anthropic","name":"Anthropic"},"importance":null},{"id":"rel_t6-BYihayp9tRzsaABpHs","title":"Claude Sonnet 4 and Claude Opus 4 models retired","path":"/release/rel_t6-BYihayp9tRzsaABpHs-claude-sonnet-4-and-claude-opus-4-models-retired","org":{"slug":"anthropic","name":"Anthropic"},"importance":null},{"id":"rel_RGLYUx7d7Li4vsN7uyQ4q","title":"ChatGPT retires GPT-5.2 models","path":"/release/rel_RGLYUx7d7Li4vsN7uyQ4q-gpt-5-2-models-retired","org":{"slug":"openai","name":"OpenAI"},"importance":null},{"id":"rel_ovBDJWt_FStOhpw_biWj8","title":"Claude Platform code execution tool now discloses 90-second cell time limit","path":"/release/rel_ovBDJWt_FStOhpw_biWj8-code-execution-tool-discloses-90-second-per-cell-time-limit","org":{"slug":"anthropic","name":"Anthropic"},"importance":2},{"id":"rel_HKLjAFmKc6Mvhsqe8UHX-","title":"Claude Platform adds response_inclusion parameter to web search and fetch tools","path":"/release/rel_HKLjAFmKc6Mvhsqe8UHX--web-search-and-fetch-tools-get-response-inclusion-parameter","org":{"slug":"anthropic","name":"Anthropic"},"importance":3},{"id":"rel_DXYaPjyG4su2fGyysx5FN","title":"Claude Platform on AWS adds GET /v1/environments/{id}/work endpoint","path":"/release/rel_DXYaPjyG4su2fGyysx5FN-self-hosted-sandbox-work-endpoint-now-available-on-aws","org":{"slug":"anthropic","name":"Anthropic"},"importance":2},{"id":"rel_pOYmtmJpaTpXCZwi7Y5Z0","title":"Claude Code v2.1.172 allows sub-agents to spawn sub-agents and fixes context issues","path":"/release/rel_pOYmtmJpaTpXCZwi7Y5Z0-sub-agents-can-spawn-sub-agents-context-and-image-errors-fixed","org":{"slug":"anthropic","name":"Anthropic"},"importance":null},{"id":"rel_hW5VT0pKzOL2qDpK0Y1O1","title":"OpenAI launches Partner Network to scale AI adoption","path":"/release/rel_hW5VT0pKzOL2qDpK0Y1O1-openai-partner-network-launches","org":{"slug":"openai","name":"OpenAI"},"importance":null},{"id":"rel_HOLmi6zZTBBzrOqy5C4ig","title":"Codex 2026-06-11 app adds rate-limit reset banking and Developer mode","path":"/release/rel_HOLmi6zZTBBzrOqy5C4ig-rate-limit-reset-banking-and-developer-mode-added","org":{"slug":"openai","name":"OpenAI"},"importance":null},{"id":"rel_R3D1VN_awz-thGK7CgeO7","title":"ChatGPT adds controls to delete memories and turn memory off","path":"/release/rel_R3D1VN_awz-thGK7CgeO7-memories-deletable-memory-toggled-off","org":{"slug":"openai","name":"OpenAI"},"importance":null}]}}