Anthropic releases API version 2026-10-21 and updated SDKs
Anthropic released API version 2026-10-21 with breaking changes to sampling parameters and message formats.
Models:Claude 4.7
On Zabio today
claude-sonnet-4-6Live. Call it by name with your Zabio key.Price and details
What changed
Anthropic released API version 2026-10-21 along with major updates to its client SDKs and the ant CLI. Requests using 2026-10-21 return a 400 error if they include temperature, top_p, or top_k, even when set to default values. Document blocks placed directly in assistant messages and non-empty citations arrays in user messages also trigger a 400 error. The new SDK and CLI versions send 2026-10-21 by default, switch the license to Apache 2.0, send API keys via Authorization: Bearer, and remove the legacy Text Completions API.
Key details
- Sending temperature, top_p, or top_k returns a 400 error under 2026-10-21.
- Placing document blocks directly in assistant messages returns a 400 error under 2026-10-21.
- Sending a text block with a non-empty citations array in a user message returns a 400 error under 2026-10-21.
- Client SDKs and the ant CLI switch licenses from MIT to Apache 2.0.
- Client SDKs and the ant CLI send API keys as Authorization: Bearer instead of X-Api-Key.
- The legacy Text Completions API is removed from TypeScript, Go, Java, and Ruby SDKs and from the ant CLI.
What it means for Zabio customers
Requests using 2026-10-21 sample using default settings, including temperature 1.0. Lowering temperature on models such as Claude Sonnet 4.6 no longer works under the new version.
Breaking changes and migration
Remove temperature, top_p, and top_k from Messages requests. Move document blocks to user messages and clear citations from user text blocks. Update proxies if they expect the X-Api-Key header.
Source
We’ve released API version 2026-10-21. Migrating to API version 2026-10-21 shows how to upgrade.
Original announcement: Claude Platform release notes
How it compares
The models in this story next to models of the same class at a similar price, all available with one Zabio key.
| Model | Input, $/M | Output, $/M | Cached, $/M | Context window | 1,000 calls | Fixing code, hard |
|---|---|---|---|---|---|---|
claude-sonnet-4-6AnthropicIn this story | 3.00 | 15.00 | 0.30 | 1,000,000 | $13.50 | 67.9 |
gpt-5.4OpenAI | 2.50 | 15.00 | 0.25 | 400,000 | $12.50 | 78.9 |
qwen3.5-omni-plusQwen | 1.40 | 8.30 | no data | 196,608 | $6.95 | no data |
gemini-4-hydrogenGoogle | 1.50 | 7.50 | no data | 200,000 | $6.75 | 78.5 |
Prices are the vendor list prices in USD per million tokens. "1,000 calls" is 1,000 requests of 2,000 input tokens and 500 output tokens each, without caching. Zabio bills the list price plus its service fee: see pricing.
Scores: percent correct for that kind of job, aggregated from public results by Epoch AI and LiveBench, read on 16 November 2026. Epoch AI: CC BY 4.0, LiveBench: Leaderboard results used with attribution; the 2.0 release states no terms of its own (read 2026-11-06); the first version's Mathematics, Language, IF results: Code Apache 2.0 and MIT; leaderboard results used with attribution
Source: Claude Platform release notes (2026-10-21)