head-to-head
| Metric | Claude Opus 5 | DeepSeek V4 Pro |
|---|---|---|
| SWE-bench Verified | 97.0% | 80.6% |
| SWE-bench Pro | — | 55.4% |
| Terminal-Bench | — | 67.9% (TB2.0) |
| Input $ / 1M | $5 | $0.435 |
| Output $ / 1M | $25 | $0.87 |
| Context | — | 1M |
| Open weights | No | Yes |
| Access | API (claude-opus-5) · Claude Code · Claude Cowork · claude.ai (Pro, Max) | Open weights (MIT) · API · self-host |
| Maker | Anthropic | DeepSeek |
what do the benchmarks actually say?
On SWE-bench Verified — real, human-validated GitHub issues resolved end-to-end — Claude Opus 5 posts 97.0% against 80.6% for DeepSeek V4 Pro, a 16.4-point gap. Verified is the closest public proxy for "can it fix a real bug in a real repo without help", which is why it anchors our ranking.
A few points either way is real but not decisive: within that band, the agent scaffolding around the model — how it retrieves files, runs tests, and retries — often matters as much as the base model. Treat the gap as a lean, not a verdict.
which is cheaper to run?
DeepSeek V4 Pro is the cheaper model: $0.435 per 1M input tokens ($0.87 output) versus $5 ($25 output) for Claude Opus 5 — roughly 11.5× less on input. Coding workloads are output-heavy — agents write diffs, tests and retries — so weight the output rate more than the input rate when you estimate a monthly bill.
DeepSeek V4 Pro ships open weights. Beyond the per-token discount, that means you can self-host it, fine-tune it, pin a snapshot so the model can't change under you, and keep code on your own infrastructure — none of which is possible with Claude Opus 5, which is API-only. For teams with compliance constraints that difference outweighs any benchmark point.
when to pick each
The highest independently measured coding score on the board at 97.0%, at half the price of Fable 5. Strongest on short and medium tasks, though GPT-5.6 Sol still edges it on multi-hour work.
The cheapest frontier-class coder — top open-weights score at ~11× less than Opus. Best pick when cost or self-hosting rules.
how were these scores verified?
We only print a number once it's confirmed against a primary source or an independent evaluation, and each row on our leaderboard records which kind it is:
- Claude Opus 5: Independent (vals.ai, observed Jul 25 2026, mini-swe-agent bash-only harness): SWE-bench Verified 97.00% ±0.76, the highest score on the board and 1st of the 75 systems vals.ai has run. Entered ranked Jul 25, 2026 after one day unranked: Anthropic published no SWE-bench Verified number at launch and still has not, so this is vals.ai's own measurement rather than a vendor claim. Read the #1 as a three-way tie, not a win — GPT-5.6 Sol is at 96.20% ±0.86 (a 0.8-point gap, ~0.7 sigma) and Claude Fable 5 at 95.00% ±0.98 (2.0 points, ~1.6 sigma), both inside the combined margin of error. We rank Opus 5 first only because it scored highest. Where the top two genuinely separate is task length, and not in Opus 5's favour: on the 1-to-4-hour tier Sol solves 98% against Opus 5's 90%, while Opus 5 leads on shorter work (98% under 15 minutes and 97% on 15-minute-to-1-hour tasks, vs 97% and 95% for Sol). Released Jul 24, 2026 at $5/$25 per 1M, the same price as Opus 4.8 and half of Fable 5. Anthropic's launch claims stay unreproducible (Frontier-Bench v0.1, CursorBench 3.2 and Zapier AutomationBench are proprietary), so this is the first externally checkable score the model has. Fast mode runs about 2.5x default speed at 2x base price.
- DeepSeek V4 Pro: Vendor-reported, as aggregated by llm-stats (June 2026). Corrected Jul 17, 2026: we previously called llm-stats an "independent tracker", but llm-stats labels its own SWE-bench Verified table "Verified: 0 / Self-reported: 104" — every score on it is vendor-claimed, so this is a vendor number, not an independent one. No independent evaluator has run V4 Pro; vals.ai has evaluated the plain DeepSeek V4 (77.4%), which is a different model. Tied with Gemini 3.1 Pro on Verified, ahead on Pro.
Full reviewsClaude Opus 5, decoded
Ranked on our AI Coding Leaderboard, updated 2026-07-27. Scores are confirmed against primary sources; prices are per 1M input tokens and can change.
- Anthropicvals.ai — SWE-bench Verified (independent) — Independent (vals.ai, observed Jul 25 2026, mini-swe-agent bash-only harness): SWE-bench Verified 97.00% ±0.76, the highest score on the board and 1st of the 75 systems vals.ai has run. Entered ranked Jul 25, 2026 after one day unranked: Anthropic published no SWE-bench Verified number at launch and still has not, so this is vals.ai's own measurement rather than a vendor claim. Read the #1 as a three-way tie, not a win — GPT-5.6 Sol is at 96.20% ±0.86 (a 0.8-point gap, ~0.7 sigma) and Claude Fable 5 at 95.00% ±0.98 (2.0 points, ~1.6 sigma), both inside the combined margin of error. We rank Opus 5 first only because it scored highest. Where the top two genuinely separate is task length, and not in Opus 5's favour: on the 1-to-4-hour tier Sol solves 98% against Opus 5's 90%, while Opus 5 leads on shorter work (98% under 15 minutes and 97% on 15-minute-to-1-hour tasks, vs 97% and 95% for Sol). Released Jul 24, 2026 at $5/$25 per 1M, the same price as Opus 4.8 and half of Fable 5. Anthropic's launch claims stay unreproducible (Frontier-Bench v0.1, CursorBench 3.2 and Zapier AutomationBench are proprietary), so this is the first externally checkable score the model has. Fast mode runs about 2.5x default speed at 2x base price.
- DeepSeekDeepSeek V4 — specs & benchmarks — Vendor-reported, as aggregated by llm-stats (June 2026). Corrected Jul 17, 2026: we previously called llm-stats an "independent tracker", but llm-stats labels its own SWE-bench Verified table "Verified: 0 / Self-reported: 104" — every score on it is vendor-claimed, so this is a vendor number, not an independent one. No independent evaluator has run V4 Pro; vals.ai has evaluated the plain DeepSeek V4 (77.4%), which is a different model. Tied with Gemini 3.1 Pro on Verified, ahead on Pro.
- BenchmarkSWE-bench — the real-GitHub-issue benchmark