Compare / head-to-head
Claude Opus 5.5vs
DeepSeek V4 Flash
Claude Opus 5.5 leads 16 of 17 shared benchmarks. Both offer a 1M-token context window.
Benchmarks from cited public sources; pricing from official pages; status from official provider feeds.
Shared benchmarks
16 – 1
Claude Opus 5.5 leads
Cheaper per token
—
list price, input + output
Larger context
Tie
both 1M tokens
Providers
2 providers
Anthropic · DeepSeek
Claude Opus 5.5
Anthropic · released 2026-09-22
textvision
DeepSeek V4 Flash
DeepSeek · released 2026-07-31
text
- Context
- 1M
- Max out
- —
- Input /1M
- —
- Output /1M
- —
- Cached /1M
- —
- Scores
- 28 · 7 core
Quality
Benchmark matrix
17 shared · 17 only Claude Opus 5.5 · 11 only DeepSeek V4 Flash| Benchmark | Claude Opus 5.5 | DeepSeek V4 Flash | Δ | Edge |
|---|---|---|---|---|
| Reported by both · 17 | ||||
| GPQA Diamond | 90.6% ↗epoch run | 91% ↗epoch run | -0.4 pt | DeepSeek V4 Flash |
| LMArena Elo | 1503.7 ↗ | 1436 ↗ | +67.7 | Claude Opus 5.5 |
| FrontierMath Tier 4 v2 (Epoch AI run) | 95% ↗epoch run | 24.4% ↗epoch run | +70.6 pt | Claude Opus 5.5 |
| FrontierMath Tiers 1-3 v2 (Epoch AI run) | 91.2% ↗epoch run | 57.5% ↗epoch run | +33.7 pt | Claude Opus 5.5 |
| LiveBench Agentic Coding (LiveBench) | 71.7% ↗ | 46.8% ↗ | +24.9 pt | Claude Opus 5.5 |
| LiveBench Coding (LiveBench) | 89.3% ↗ | 75% ↗ | +14.3 pt | Claude Opus 5.5 |
| LiveBench Data Analysis (LiveBench) | 80.3% ↗ | 79.3% ↗ | +1 pt | Claude Opus 5.5 |
| LiveBench Instruction Following (LiveBench) | 67% ↗ | 65.5% ↗ | +1.5 pt | Claude Opus 5.5 |
| LiveBench Language (LiveBench) | 86.3% ↗ | 79.2% ↗ | +7.1 pt | Claude Opus 5.5 |
| LiveBench Mathematics (LiveBench) | 97.1% ↗ | 86.8% ↗ | +10.3 pt | Claude Opus 5.5 |
| LiveBench Reasoning (LiveBench) | 92.2% ↗ | 86.6% ↗ | +5.6 pt | Claude Opus 5.5 |
| LMArena WebDev (LMArena) | 1815.4 ↗ | 1581 ↗ | +234.4 | Claude Opus 5.5 |
| Mystery Game Puzzles (Epoch AI run) | 71% ↗epoch run | 34% ↗epoch run | +37 pt | Claude Opus 5.5 |
| OTIS Mock AIME 2024-2025 (Epoch AI run) | 100% ↗epoch run | 94.4% ↗epoch run | +5.6 pt | Claude Opus 5.5 |
| SimpleBench (SimpleBench) | 88.4% ↗ | 61.1% ↗ | +27.3 pt | Claude Opus 5.5 |
| SimpleQA Verified | 72.2% ↗epoch run | 33.6% ↗epoch run | +38.6 pt | Claude Opus 5.5 |
| Terminal-Bench 4.0 (Vals AI) | 65.2% ↗ | 18.7% ↗ | +46.5 pt | Claude Opus 5.5 |
| Only Claude Opus 5.5 reports · 17 | ||||
| APEX-Agents (Mercor) | 73.5% ↗ | not reported | — | — |
| AutomationBench 1.0.6 | 40% ↗ | not reported | — | — |
| Chartography (with tools) | 89% ↗ | not reported | — | — |
| CursorBench 4.0 | 57.8% ↗ | not reported | — | — |
| EBR-bench (Epoch AI run) | 71.4% ↗epoch run | not reported | — | — |
| FrontierCode v1.1 (Main) | 54.4% ↗ | not reported | — | — |
| FrontierSWE V2 (Proximal Labs) | 62.3% ↗ | not reported | — | — |
| Furniture Assembly (Epoch AI run) | 83.3% ↗epoch run | not reported | — | — |
| GDPval-AA v2.1 | 1846 ↗ | not reported | — | — |
| Humanity's Last Exam (with tools) | 67.7% ↗ | not reported | — | — |
| LMArena Agent (LMArena) | 0.1382 ↗ | not reported | — | — |
| MirrorCode (Epoch AI run) | 77.4% ↗epoch run | not reported | — | — |
| OSWorld 2.0 (partial) | 81.8% ↗ | not reported | — | — |
| SAGE (Vals AI) | 45.8% ↗ | not reported | — | — |
| Terminal-Bench 4.0 | 66.4% ↗ | not reported | — | — |
| Terminal-Bench Science 0.1 | 58.7% ↗ | not reported | — | — |
| Vending-Bench 2 (Andon Labs) | 9235.25 ↗ | not reported | — | — |
| Only DeepSeek V4 Flash reports · 11 | ||||
| Agents' Last Exam | not reported | 25.2% ↗ | — | — |
| AutomationBench Public | not reported | 25.1% ↗ | — | — |
| Chess Puzzles (Epoch AI run) | not reported | 33% ↗epoch run | — | — |
| Cybergym | not reported | 76.7% ↗ | — | — |
| DeepSWE | not reported | 54.4% ↗ | — | — |
| DeepSWE v1.1 (Datacurve) | not reported | 53.3% ↗ | — | — |
| NL2Repo | not reported | 54.2% ↗ | — | — |
| Terminal Bench 2.1 | not reported | 82.7% ↗ | — | — |
| Toolathlon-Verified | not reported | 70.3% ↗ | — | — |
| Toolathlon-Verified (HKUST) | not reported | 70.7% ↗ | — | — |
| WeirdML (Håvard Tveit Ihle) | not reported | 63% ↗ | — | — |
Scores tagged "epoch" or "matharena" are independent runs, used only where the lab has not published its own; ⚠ marks rows MathArena flags as released after the competition. Higher is better on every row. Δ is Claude Opus 5.5 minus DeepSeek V4 Flash in the benchmark's own unit. "Not reported" means the lab has not published that figure; it is not a zero. ↗ opens the source.
Specs & pricing
Side by side
Official pricing pages and model cards| Spec | Claude Opus 5.5 | DeepSeek V4 Flash | Edge |
|---|---|---|---|
| Context window | 1M tokens | 1M tokens | Tie |
| Max output | 128K tokens | — | — |
| Input price / 1M | $4 ↗ | — | — |
| Output price / 1M | $20 ↗ | — | — |
| Cached input / 1M | $0.2 ↗ | — | — |
| Input + output / 1M Lower is cheaper. List prices; batch, tool and regional fees excluded. | $24 | — | — |
| Modalities | text · vision | text | Claude Opus 5.5 |
| Released | 2026-09-22 | 2026-07-31 | — |
| Cited benchmark scores | 34 | 28 | — |
Reliability
Provider status
Live from /statusMore matchups
Claude Opus 5.5 vs …
Models sharing the most benchmarksMore matchups
DeepSeek V4 Flash vs …
Models sharing the most benchmarksBuilt by Respan
Which one wins on your data?
Public benchmarks are a starting point. Run Claude Opus 5.5 and DeepSeek V4 Flash on your own prompts with Respan evals, or route to either through one gateway key with automatic failover.