Model components / Fine-tuned / chat models

Claude Opus 4.x

Anthropic's heavy-reasoning Opus tier, tuned for sustained agentic coding and long-horizon planning across large codebases. Opus 4 scored 72.5% on SWE-bench Verified at release (May 2025 era); the line has iterated through Opus 4.5 (80.9%), Opus 4.6 (80.8%), and Opus 4.7 (April 2026), which hits 87.6% SWE-bench Verified and 98.5% computer-use visual acuity. Opus 4.6 and 4.7 are tied at the top of LMArena (Elo ~1500). A leaked 'Claude Mythos Preview' tops SWE-bench at 93.9% and is presumed to be the next-generation Opus / Claude 5 candidate.

Anthropic Claude Opus 4.x flagship line: Opus 4 (May 22 2025) -> 4.5 (Nov 2025) -> 4.6 (Feb 2026) -> 4.7 (Apr 16 2026) -> 4.8 (May 28 2026, current frontier, model id claude-opus-4-8). Post-trained chat/agentic flagship, closed/API + Bedrock + Vertex + claude.ai. Note: Opus 4.7 is a frozen base_pretrained anchor; this record scores the Opus 4.x line in finetuned_chat. Verified via Anthropic Opus 4.8 + Claude 4 pages June 2026. Consolidated on 2026-07-29 from claude-opus-4-7; openness follows claude-opus-4-x, the release that currently governs. See sources/slug_aliases.yaml.

Openness

1 high confidence

1.0

weights: closed
data: closed
code: closed
license: Proprietary(API/Bedrock/Vertex/claude.ai)

No weights distributed in any form. Served only through the Anthropic API, Bedrock, Vertex and claude.ai, so 1/closed on every dimension. Opus 4.x governs as the current release; 4.7 scored identically, and nothing about the tier's openness turns on which release is read. Consolidated because Anthropic markets Claude Opus as the product and 4.7 and 4.8 are versions of it.

https://www.anthropic.com/news/claude-opus-4-8 recorded 2026-06-04
Opus 4.8 (May 28 2026) proprietary, claude-opus-4-8 via API at $5/$25 per Mtok
https://www.anthropic.com/news/claude-4 recorded 2026-06-04
Opus 4 line origin (May 22 2025), proprietary, API/Bedrock/Vertex

Adoption

4 medium confidence

4.0

Directional market-position estimate (not a measured usage figure). Opus is Anthropic's premium tier top-tier reach but lower volume than the Sonnet workhorse, so 4 rather than 5. Anthropic's launch post lists 28 enterprise testimonials (Stripe, Replit, Vercel, etc.); Opus flagships appear on public LLM marketplace leaderboards (e.g. OpenRouter). Overrides the prior frozen-anchor convention (held null), which systematically understated dominant closed incumbents and could mask the openness gap.

https://www.anthropic.com/news/claude-opus-4-7 recorded 2026-06-04
28 enterprise testimonials; no quantitative per-model usage figures
https://openrouter.ai/rankings recorded 2026-06-23
public per-model token-share leaderboard where Claude flagships rank among the most-used models

Capability

5 high confidence

5.0

Anthropic-reported: state-of-the-art on GDPval-AA and Finance Agent; CursorBench 70% (vs 58% for Opus 4.6); BigLaw Bench 90.9% at high effort; gains on SWE-bench Verified/Pro/Multilingual. Independent Gemini 3.5 Flash comparisons place Opus 4.7 in the frontier field. Vendor-reported.

https://www.anthropic.com/news/claude-opus-4-7 recorded 2026-06-04
flagship phase-C verification source
https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-5/ recorded 2026-06-04
flagship phase-C verification source

Unchanged since 2026-07-29 (last edited, not re-checked)