InfraAnthropicConfirmed
Story 5 of 9 in the Monday 2 November editionAnthropic adds server-side message threads to cut long-context Claude API latency
Launched in public beta on 2 November 2026, message threads keep conversation state in encrypted server memory for five minutes so subsequent turns transmit only new messages and cut median time to first token by roughly 35%.

Anthropic launched server-side message threads in public beta on the Claude Messages API (beta header message-threads-2026-08-12) on 2 November 2026. Long-running coding and tool-use agents typically re-upload hundreds of thousands of conversation, system and tool-schema tokens on every turn, creating network upload, JSON parsing and validation overhead before inference even starts.
Creating a server-side thread stores the conversation history, system prompt and tool definitions in encrypted server RAM without writing them to disk. Follow-up calls pass previous_message_id and transmit only the new user or tool-result messages, supporting up to four cache_control breakpoints. Thread state expires after a rolling five-minute idle window and is permanently deleted within one hour; if a thread expires, the API returns a 404 thread_not_found error and Anthropic’s updated SDKs automatically replay the full local history.
Key numbers
- Roughly 35% lower median time to first token on 500k to 1M token requests.
- Rolling 5-minute idle expiry in encrypted RAM, with hard deletion within 1 hour and zero disk writes.
- Up to 4 cache_control breakpoints per threaded request.
- Standard Messages API per-token rates apply, with Zero Data Retention and HIPAA workloads excluded.
Sources
- PrimaryCut latency on long-context API requests with message threadsclaude.com
- OfficialMessage threads documentation on Claude Platformplatform.claude.com
- OfficialClaude API pricingclaude.com
- On XClaude message threads announcement on Xx.com
The rest of the day
Read the whole day01Coding agentDeepSeek
DeepSeek Harness adds Creator Mode for building live UI plugins in chat
Announced on 2 November 2026, the fourth session mode in DeepSeek’s MIT-licensed desktop and local web agent harness generates, lints, themes and hot-installs custom interface plugins inside an active workspace.
02Open sourceDeep Principle
Deep Principle releases Hydrogen-1.0 open-weight model for chemistry and materials
Built on Z.ai’s GLM-5.3 and released under the MIT licence on 2 November 2026, Hydrogen-1.0 pairs laboratory notebook trajectories with reinforcement learning across molecular, materials and life-science workflows.
03ResearchOpenAI
OpenAI introduces LifeVal to test AI agents on everyday household and civic tasks
Grounded in government time-use diaries and a survey of 3,995 people, the evaluation published on 2 November 2026 measures browser use, computer use and connectors across seven personal-life domains.
04ModelMeituan
Meituan launches LongCat-2.5 with 1.6T parameters and discrete visual tokens
Detailed in a 1 November 2026 technical report and opened on the LongCat API on 2 November 2026, the Mixture-of-Experts model activates roughly 48B parameters per token across text and native-resolution images.
Spotted an error? Email aiforsuccess.in+compass@gmail.com or tag @ai_for_success on X







