Claude Opus 4.8 vs Claude Opus 4.7
Anthropic released Claude Opus 4.8 on May 28, 2026, just six weeks after Opus 4.7. Pricing held steady at $15 input and $75 output per million tokens, and the 1 million token context window carries over unchanged. The gains are in capability and economics: Anthropic's own numbers put agentic coding at 69.2% (up from 64.3%), knowledge work at 1890 (up from 1753), and agentic financial analysis at 53.9% (up from 51.5%), while the new fast mode runs roughly 2.5x quicker and about three times cheaper than before. Anthropic also frames 4.8 as more honest about its own progress and better at working independently for longer. Here is the full generational diff.
Head-to-Head Specs
| Spec | Claude Opus 4.8 | Claude Opus 4.7 |
|---|---|---|
| Provider | Anthropic | Anthropic |
| Input Price | $5.00/1M | $15.00/1M |
| Output Price | $25.00/1M | $75.00/1M |
| Context Window | 1M | 1M |
| Released | 2026-05 | 2026-04 |
| Capabilities | text, vision, tool-use, code | text, vision, tool-use, code |
Benchmark Scores
| Benchmark | Claude Opus 4.8 | Claude Opus 4.7 | Winner |
|---|---|---|---|
| SWE-bench | 88.6 | 87.6 | Claude |
| GPQA Diamond | 93.6 | 76.5 | Claude |
See the full benchmark leaderboard for all models.
Category Breakdown
Opus 4.8 scores 69.2% vs 4.7 at 64.3% on Anthropic's agentic coding measure
Opus 4.8 posts 1890 vs 4.7 at 1753 on Anthropic's knowledge work score
Opus 4.8 reaches 57.9% vs 4.7 at 54.7% on multidisciplinary reasoning with tools
Opus 4.8 reaches 83.4% vs 4.7 at 82.8%
Opus 4.8 improves to 53.9% from 4.7 at 51.5%
Opus 4.8 fast mode is roughly 2.5x faster and about 3x cheaper than the previous fast mode
Both cost $15 input and $75 output per 1M tokens
Both ship a 1 million token context window
Choose Claude Opus 4.8 when:
- ▸Agentic coding and long-running asynchronous workflows
- ▸Tasks that benefit from higher effort settings (extra or max)
- ▸Workloads where the cheaper, faster fast mode lowers cost
- ▸Teams that want the most honest progress reporting
Choose Claude Opus 4.7 when:
- ▸Existing prompts and evals already tuned to 4.7 behavior
- ▸Version-pinned enterprise contracts
- ▸Regression testing against the prior generation
- ▸Workloads where 4.7 already meets quality and latency targets
Frequently Asked Questions
Which is better, Claude Opus 4.8 or Claude Opus 4.7?
It depends on your use case. Claude Opus 4.8 from Anthropic excels at agentic coding and long-running asynchronous workflows, while Claude Opus 4.7 from Anthropic is better for existing prompts and evals already tuned to 4.7 behavior. See the full comparison above for detailed benchmarks and pricing.
How much does Claude Opus 4.8 cost compared to Claude Opus 4.7?
Claude Opus 4.8 costs $5.00 input and $25.00 output per 1M tokens. Claude Opus 4.7 costs $15.00 input and $75.00 output per 1M tokens.
What is the context window difference between Claude Opus 4.8 and Claude Opus 4.7?
Claude Opus 4.8 supports 1M tokens, while Claude Opus 4.7 supports 1M tokens.