Skip to content

feat(models): update the capable-tier Sonnet to Claude Sonnet 5#1198

Merged
silversurfer562 merged 1 commit into
mainfrom
feat/sonnet-5-model-update
Jul 1, 2026
Merged

feat(models): update the capable-tier Sonnet to Claude Sonnet 5#1198
silversurfer562 merged 1 commit into
mainfrom
feat/sonnet-5-model-update

Conversation

@silversurfer562

Copy link
Copy Markdown
Member

What

Updates attune's default "capable" model from claude-sonnet-4-6 to the newly released claude-sonnet-5 (Claude Sonnet 5, GA 2026-06-09).

I verified the model ID against the live Anthropic models doc before touching anything — my cached reference predated the release, so this confirms claude-sonnet-5 is real and current (Sonnet 4.6 is now legacy).

Scope — full swap (src/ + tests/)

A clean ID swap claude-sonnet-4-6 → claude-sonnet-5 across 65 source/test files (169 occurrences), plus the CHANGELOG. Pricing is unchanged ($3/$15 per MTok standard), so no pricing math moved.

  • Registry CAPABLE ModelInfo + TIER_PRICING comment
  • Adaptive routing "capable" map, executor, token estimator, telemetry baselines
  • All scattered literal model_id="claude-sonnet-4-6" defaults across providers, configs, workflows, agent factory, orchestration tools
  • Matching test expectations updated so CI stays green
  • Stale # Sonnet 4.6 prose comments refreshed to # Sonnet 5
  • Left untouched: claude-sonnet-4-5 (legacy 4.5 — a different model) and the historical CHANGELOG entry that introduced 4.6

Sonnet 5 facts (from the live doc)

Model ID claude-sonnet-5 (Bedrock anthropic.claude-sonnet-5)
Pricing $3 / $15 per MTok standard (intro $2/$10 until 2026-08-31)
Context / output 1M / 128K
Thinking adaptive; effort defaults to high

Verification

  • 962 model/routing/pricing/telemetry/provider tests pass (tests/models/, tests/unit/models/, tests/routing/, pricing lookup, telemetry analytics, retired-snapshot guard).
  • Registry resolves: get_model('anthropic','capable').id → claude-sonnet-5.
  • The only failures are the live-API integration tests in tests/models/test_sonnet_opus_fallback.py (@pytest.mark.integration, skipif(not HAS_API_KEY)) — a local creditless-key 400, not the swap. CI runs these with ANTHROPIC_API_KEY="" so they're skipped. The new ID was accepted by the API (billing 400, not a 404 model-not-found), which itself confirms the ID is valid.
  • black + ruff clean on all changed files.

Not in this PR (follow-up if wanted)

claude-sonnet-4-6 also appears in docs/ (6), website/ (7), examples/ (1), and .claude/ (3) — out of the chosen src+tests scope. Happy to sweep those in a follow-up (website has its own deploy + content-accuracy guard).

🤖 Generated with Claude Code

Swaps the default "capable" model from claude-sonnet-4-6 to the newly
released claude-sonnet-5 (GA 2026-06-09) across the registry, adaptive
routing, providers, telemetry, token estimation, and the literal
model_id defaults scattered through the codebase. Verified the ID
against the live Anthropic models doc before changing anything.

Pricing is unchanged: $3/$15 per MTok standard (introductory $2/$10
through 2026-08-31). 1M context, 128K max output, adaptive thinking,
effort defaults to high. Sonnet 4.6 is now a legacy model.

- Full claude-sonnet-4-6 -> claude-sonnet-5 swap across src/ + tests/
  (65 files); matching test expectations updated so CI stays green.
- Registry CAPABLE ModelInfo + TIER_PRICING comment refreshed.
- Stale "Sonnet 4.6" prose comments updated to "Sonnet 5".
- Left claude-sonnet-4-5 (legacy 4.5) references untouched.
- CHANGELOG Unreleased "Changed" entry added.

962 model/routing/pricing/telemetry/provider tests pass. The only
failures are the live-API integration tests in
tests/models/test_sonnet_opus_fallback.py (skipif HAS_API_KEY) — a
local creditless-key 400, skipped in CI; the new ID is accepted (no
404), confirming the swap.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
@vercel

vercel Bot commented Jun 30, 2026

Copy link
Copy Markdown
Contributor

The latest updates on your projects. Learn more about Vercel for GitHub.

Project Deployment Actions Updated (UTC)
attune-ai.dev Ready Ready Preview, Comment Jun 30, 2026 11:51pm
website Ready Ready Preview, Comment Jun 30, 2026 11:51pm

@codecov

codecov Bot commented Jun 30, 2026

Copy link
Copy Markdown

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

@silversurfer562 silversurfer562 merged commit 59eaab8 into main Jul 1, 2026
50 of 51 checks passed
@silversurfer562 silversurfer562 deleted the feat/sonnet-5-model-update branch July 1, 2026 00:00
silversurfer562 added a commit that referenced this pull request Jul 1, 2026
Follow-up to the src/tests Sonnet 5 swap (#1198): updates the
claude-sonnet-4-6 references in served docs, the example config, and the
blog tutorials to claude-sonnet-5, so user-facing surfaces match the
registry default.

- docs/ served pages (installation model list, agent-factory capable
  table, API_REFERENCE + agent-factory-readme code examples)
- examples/config/attune.config.example.yml
- website/content/blog/*.mdx code snippets

Left untouched (historical / generated, not the cleanup target):
.claude/lessons.md (illustrative stable-alias example), .claude/plans/*
(dated planning docs), docs/archive/**, weekly-reports/**, the CHANGELOG
entry that introduced 4.6, and website/public/framework-docs/** (built
mkdocs HTML — regenerated from docs/ at build time).

Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

cli core documentation Improvements or additions to documentation tests

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant