Start with the technical decision in front of you, then inspect the evidence, limitations and implementation guidance behind it.
Each desk connects evidence to a build, buy, deploy, evaluate or stop decision.
Search across titles, summaries, evidence status and tags, or narrow the library to one decision desk.
Gemini Decision 1 scores up to 1,000 questions against shared input, with no generated output tokens
Google’s November 11 launch brings a dedicated decision endpoint to the Gemini Developer API. It costs $0.10 per million input tokens, supports text and media, and returns probabilities rather than prose—but the headline latency claim has a narrower test envelope than the full API limits.
The million-token model has global, US and India routing, but the model card flags important Converse and reasoning differences
AWS’s November 10 guide explains how to use Kimi K3.1 Pro Highspeed through Bedrock’s OpenAI-compatible APIs. Global Standard pricing is $4 input and $20 output per million tokens; residency, caching and API choice change the deployment decision.
The Qwen-based agent model has safetensors and three GGUF variants, including a roughly 0.53 GB download
Tether AI Research announced BotPsy on November 11. Public weights and Apache 2.0 licensing make it a concrete local-agent candidate, but its own results show weak deep research and do not establish phone performance.
The iPhone rollout adds conversational AI to the dashboard, with Gmail and Calendar access subject to setup and eligibility
Google began rolling out Gemini on Apple CarPlay on November 11. The app supports voice conversations, recent chats and iPhone history sync, but global rollout does not mean every personalization feature is available everywhere.
Its November 11 guide starts with native AWS access and adds gateways only when routing or enforcement requires them
AWS’s new rollout guide connects workstation-based Codex to supported OpenAI models on Bedrock using IAM Identity Center, with AgentCore and LLM gateways as alternatives. It is deployment guidance, not a new Codex model or proof that every AWS account has access.
InvokeWebSearch brings Amazon’s index to custom agent loops in six Regions at $7 per 1,000 queries
AWS’s November 11 launch lets applications call AgentCore Web Search directly with IAM credentials. The API removes gateway setup and charges, but leaves retries, caching and request-level source filters to application code.
The November 9 release cuts standard token prices, adds optional fast mode and reaches AWS, Google Cloud and Azure; Mythos 5.2 remains gated
Claude Fable 5.2 is available as claude-fable-5-2 for demanding coding and technical work. Anthropic reports faster generation and stronger evaluations, but route-specific retention and effort settings still matter.
Thinking Machines Lab’s November 5 model now has a zero-priced API listing; promised open weights are a separate milestone
OpenRouter lists thinkingmachines/inkling-2-small:free with image inputs, reasoning and a 1,048,576-token context. Thinking Machines positions the 17B-active model for coding agents, but calls out tool-selection and chat limitations.
A November 5 study classifies 207,049 consumer-plan research sessions and finds heavy use in computational fields—but its 2× to 16× acceleration figures are modeled scenarios, not observed productivity gains
Anthropic analyzed 207,049 research sessions from Claude and Claude Code. Data analysis and writing dominate current use, while the study’s larger acceleration figures depend on strong automation and cost assumptions.
The November 6 study reports profitable trading against replayed game data and capable fleet coordination, but directional bets were weak and warned teams almost never identified a prompted saboteur
Anthropic put Claude agents into EVE Online market replays and live EVE Frontier missions. The first-party results show disciplined execution and coordination alongside brittle adaptation, weak forecasting and a serious multi-agent oversight gap.
Real-time inference-component endpoints can now split A100, H100 and supported Blackwell GPUs into hardware-isolated slices, but layouts are fixed at endpoint creation and customers still pay for the full instance
AWS launched managed GPU partitioning for SageMaker AI on November 6. It can place multiple models on one accelerator without self-managing MIG, but support is limited to inference components, four instance families and seven Regions.
The November 6 engineering report separates transcripts, notes, navigable representations and injected context, then tests offline “dreaming” passes that correct notes and build wikis—but the results remain first-party and task-specific
Anthropic published new agent-memory evaluations on November 6. Its strongest practical lesson is not to inject everything: preserve transcripts, keep a small index, validate notes against source records and test whether a maintained representation beats search.
The 1.73B-parameter text model is callable as liquid/lfm-3-1.7b or liquid/lfm-3-1.7b:free with 262K context, mandatory reasoning and an 8,192-token output ceiling; the provider may retain prompts and outputs for training
OpenRouter added free and paid LFM3-1.7B routes on November 6. The free endpoint makes the compact agent model easy to trial, but its data-use terms and small-model limits rule out sensitive or high-stakes workloads without further review.
The cyber-specialized GPT-6 Astra variant is initially available through the API and Codex to selected approved customers, with no public model ID or price announced
OpenAI released GPT-6 Cyber on November 5 for a select group of approved Daybreak Red customers. The company reports large gains in exploit development and scope adherence, but access remains restricted and the results are primarily vendor evaluations.
The generally available gemini-omni-1.2-flash API adds longer native video, multi-turn editing and stronger reference controls, while existing Omni 1.1 calls will reroute on December 3
Google released Gemini Omni 1.2 Flash on November 5. The paid API model generates and edits video up to 15 seconds with native audio, costs about $0.07 per second at 720p, and replaces Omni 1.1 automatically in December.
The November 5 API routes choose models per task and session, charge the selected model’s rate without a router surcharge, and leave quality, price and capabilities dependent on the downstream choice
OpenRouter released openrouter/auto for general workloads and openrouter/auto-code for coding agents on November 5. The routes can change models across tasks, expose the selected model in responses, and inherit variable model pricing and limits.
The 65B dense coding model is available in BF16 and NVFP4 under Apache 2.0, while its headline agentic scores and $32 million development-cost figure remain vendor-reported
Ricursive released Ripple-65B Preview on November 4 as a 65B dense, Apache-licensed model with downloadable BF16 and NVFP4 checkpoints. OpenRouter lists a 1M-context route at $0.20 input and $0.60 output per million tokens.
Hugging Face’s November 5 technical review puts new attention on a compact speech family first released October 20—but its licenses, access gates and automated leaderboard results differ by checkpoint
Gradium’s Phonon family offers 31M and 40M text-to-speech weights for local streaming across five languages. The English 31M checkpoint is Apache 2.0; the larger variants are non-commercial, and the new leaderboard claims still need workload-specific listening tests.
The November 4 API listing exposes a Tencent-hosted FP4 endpoint at a promotional $0.417/$1.2505 per million tokens, but no working first-party model card or benchmark report was available during review
OpenRouter added Tencent Hy4 on November 4 with text-and-image input, a one-million-token context window and tool support. The endpoint is callable now, while architecture, training and benchmark evidence remain sparse.
The anonymous 1M-context model produced a polished landing page, a working kanban board and a near-copy of a screenshot at zero cost, with slow responses, small chart bugs and a data-retention catch
Crouton Alpha appeared on OpenRouter on November 3 as a free “stealth” model from an unnamed lab. AccessAllGPT sent it five front-end tasks and rendered every result offline in Chrome. All five ran without page errors; the screenshots show what it does well and where it slips.
The 2.8-trillion-parameter multimodal MoE is downloadable now, while its custom license and unusually large serving footprint make “open weight” different from “easy to self-host”
Moonshot AI’s Kimi K3.1 Pro repository went public on November 2 with model weights, code and a custom license. Two managed API routes are also live, but buyers should separate vendor benchmark claims from deployability.
The two API models are live on the Images endpoints with complexity controls and identical token rates, while OpenAI’s model pages currently mark Responses API support unavailable
OpenAI announced Images 2.6 on November 4 across ChatGPT, Codex and the API. Builders get two named API models, animated GIF and static SVG output, markup-guided edits and a new complexity control—but several endpoint and output constraints matter.
Gemini 3.8 Flash, Claude Opus 5.5 and GPT-6 Sol and Luna fell to 13% accuracy or lower on the longest general-finite cases, while the open benchmark separates four kinds of state updates
Ai2 released StateBench, its code and data on November 3. The benchmark tests whether language models can preserve changing entity state across long inputs—and shows that strong frontier models still break down on the hardest update chains.
The October 29 specification adds explicit rules for oversight, contextual memory, real-world ties, side effects and teen safety—but OpenAI says production models do not yet fully match it
OpenAI has reorganized its public Model Spec around faithfulness, harm reduction, honesty, human agency and care. The document now makes context use and user control more explicit, while stopping short of announcing a model, API, price or completed behavior rollout.
