AI for Devs covers model releases, coding agents, MCP, inference engines, and LLM APIs for the engineers who build on them. News briefs link to primary sources, articles show the mechanism with diagrams and code, and open datasets track prices and model retirements.
Every news brief links to the announcement, paper, or changelog it came from. Every deep dive explains the mechanism underneath, shows the trade-off, and says who should use it. If a claim has no source we can link, it does not run.
The claude-fable-5-2 model costs $8 input and $40 output per million tokens, down from $10 and $50 for Fable 5.1, with a 1M-token context and 128K max output. Before you switch, check any code that edits history. Changing the system prompt, tools, or earlier messages and then replaying a later thinking block returns a 400 on every account.
It costs $0.10 per million input tokens and $0.50 per million output tokens, with a 1M-token context window and up to 512K output tokens. Meta says it matches Claude Haiku 5.5's best Artificial Analysis score at about a third of the cost, though Meta's chart labels Ember's per-task costs as estimates at proposed pricing.
Fast mode for Claude Fable 5.2 costs $16 per million input tokens and $80 per million output tokens on the Claude API. Anthropic reports up to 2.5x higher output tokens per second. You turn it on by setting speed: "fast" on a request, with no beta header.
GitHub secret scanning now scans GitHub Actions workflow logs, in public preview for GitHub Enterprise Cloud with Secret Protection or Advanced Security. Enterprise owners turn on the policy, which is off by default. Only runs that finish after you enable it are scanned, and preview alerts send no notifications.
You can now manage code review settings through the APIs from enterprise policy down to individual repositories, and see which skills, MCP servers, and custom instructions Copilot reported as sources. Enterprise and organization endpoints don't support GitHub Apps or fine-grained personal access tokens; repository endpoints support both.
As of November 6, 2026, Gemini API requests to gemini-embedding-2-preview route to gemini-embedding-2 and are billed at Gemini Embedding 2 rates. Google says the stable model uses the same API schema, pricing, and embedding space, so you only need to update the model string, with no vector re-indexing.
An agent decides what to ask the model and which tools to call.
A KV cache holds attention keys and values during LLM decode. Size one from config.json, see what concurrency costs, and how prompt caching reuses it.
11 min readInference and serving
Deep dive15 min readCoding agents
Guide14 min readAPIs and pricing
Guide16 min readAPIs and pricing
Deep dive12 min readEvals and observability
Guide11 min readEvals and observability
1,433rates
Input, cached, and output prices for 495 models from 80 sellers.
DatasetCSV · JSON · Checked Nov 10 2026
385models
Deprecation status, shutdown dates, and replacements. 58 retire in the next 90 days.
DatasetCSV · JSON · Checked Nov 10 2026
901models
Text and multimodal models from 24 labs, with license, size, and downloads.
DatasetCSV · JSON · Checked Nov 10 2026
62rates
On-demand and spot prices from 7 clouds, per GPU-hour. H100 from $2.89/hr.
DatasetCSV · JSON · Checked Nov 10 2026
47platforms
Model APIs, inference engines, coding agents, agent frameworks, and vector databases.
DirectoryBy category
54terms
KV cache, MCP, RAG, speculative decoding, quantization, and tool calling, defined plainly.
GlossaryA to Z
The RSS feed carries every news brief and article with its date and source. No account, no tracking pixel.