Compare
Four models, one page
Pick the models you are actually choosing between. Every row is scaled within itself, so a bar means something next to its neighbours and nothing across rows.
| Specification | V3.2 ExpDeepSeek | Claude Opus 5Anthropic |
|---|---|---|
| Released | Sep 29, 2025 | Jul 24, 2026 |
| Context window | 164K | 1M |
| Max output | 66K | 128K |
| Input / 1M | $0.27 | $5 |
| Output / 1M | $0.41 | $25 |
| Cost per task | $0.05 | $2.25 |
| Arena Elo | — | 1,393 |
| Serving providers | 3 | 5 |
| Parameters | 685.4B | Undisclosed |
| Licence | mit | Proprietary |
| Capabilities |
|
|
Metrics side by side
Each row is scaled to the largest value in that row — bars compare within a row, never across rows
- V3.2 Exp
- Claude Opus 5
Intelligence Index
V3.2 Expnot measuredClaude Opus 563.1Coding Index
V3.2 Expnot measuredClaude Opus 578.0Agentic Index
V3.2 Expnot measuredClaude Opus 559.2Output speed
V3.2 Exp22 t/sClaude Opus 587 t/sContext window
V3.2 Exp164KClaude Opus 51MLatency · lower is better
V3.2 Exp1.64sClaude Opus 51.12sBlended price / 1M · lower is better
V3.2 Exp$0.305Claude Opus 5$10Cost per task · lower is better
V3.2 Exp$0.05Claude Opus 5$2.25
Rows marked “lower is better” still draw a longer bar for a larger number — read the value, not just the length. Arena Elo is in the specification table above instead: it has no meaningful zero, so a bar would flatten the gaps.
View as table
| Metric | V3.2 Exp | Claude Opus 5 |
|---|---|---|
| Intelligence Index | — | 63.1 |
| Coding Index | — | 78.0 |
| Agentic Index | — | 59.2 |
| Output speed | 22 t/s | 87 t/s |
| Context window | 164K | 1M |
| Latency · lower is better | 1.64s | 1.12s |
| Blended price / 1M · lower is better | $0.305 | $10 |
| Cost per task · lower is better | $0.05 | $2.25 |
Evaluation scores
Percentage correct on a common 0–100% scale
- V3.2 Exp
- Claude Opus 5
GPQA Diamond
V3.2 Expnot measuredClaude Opus 593.2%Humanity's Last Exam
V3.2 Expnot measuredClaude Opus 554.9%SciCode
V3.2 Expnot measuredClaude Opus 555.7%τ²-bench
V3.2 Expnot measuredClaude Opus 542.1%AA-LCR long context
V3.2 Expnot measuredClaude Opus 575.7%
A missing bar means that evaluation was not run for that model — it is not a zero.
View as table
| Evaluation | V3.2 Exp | Claude Opus 5 |
|---|---|---|
| GPQA Diamond | — | 93.2% |
| Humanity's Last Exam | — | 54.9% |
| SciCode | — | 55.7% |
| τ²-bench | — | 42.1% |
| Terminal-Bench Hard | — | — |
| LiveCodeBench | — | — |
| AA-LCR long context | — | 75.7% |
| AIME 2025 | — | — |
Estimated from list pricing: 50K input tokens plus 80K output tokens for reasoning models (25K for non-reasoning).