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 | V3.1 TerminusDeepSeek | Claude Opus 5Anthropic |
|---|---|---|
| Released | Sep 22, 2025 | Jul 24, 2026 |
| Context window | 164K | 1M |
| Max output | 33K | 128K |
| Input / 1M | $0.27 | $5 |
| Output / 1M | $1 | $25 |
| Cost per task | $0.09 | $2.25 |
| Arena Elo | — | 1,393 |
| Serving providers | 5 | 5 |
| Parameters | 684.5B | 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
- V3.1 Terminus
- Claude Opus 5
Intelligence Index
V3.1 Terminus21.7Claude Opus 563.1Coding Index
V3.1 Terminusnot measuredClaude Opus 578.0Agentic Index
V3.1 Terminusnot measuredClaude Opus 559.2Output speed
V3.1 Terminus55 t/sClaude Opus 587 t/sContext window
V3.1 Terminus164KClaude Opus 51MLatency · lower is better
V3.1 Terminus803msClaude Opus 51.12sBlended price / 1M · lower is better
V3.1 Terminus$0.453Claude Opus 5$10Cost per task · lower is better
V3.1 Terminus$0.09Claude 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 | V3.1 Terminus | Claude Opus 5 |
|---|---|---|
| Intelligence Index | 21.7 | 63.1 |
| Coding Index | — | 78.0 |
| Agentic Index | — | 59.2 |
| Output speed | 55 t/s | 87 t/s |
| Context window | 164K | 1M |
| Latency · lower is better | 803ms | 1.12s |
| Blended price / 1M · lower is better | $0.453 | $10 |
| Cost per task · lower is better | $0.09 | $2.25 |
Evaluation scores
Percentage correct on a common 0–100% scale
- V3.1 Terminus
- Claude Opus 5
GPQA Diamond
V3.1 Terminus75.1%Claude Opus 593.2%Humanity's Last Exam
V3.1 Terminus8.7%Claude Opus 554.9%SciCode
V3.1 Terminus32.1%Claude Opus 555.7%τ²-bench
V3.1 Terminus37.1%Claude Opus 542.1%Terminal-Bench Hard
V3.1 Terminus31.8%Claude Opus 5not measuredLiveCodeBench
V3.1 Terminus52.9%Claude Opus 5not measuredAA-LCR long context
V3.1 Terminus45.0%Claude Opus 575.7%AIME 2025
V3.1 Terminus53.7%Claude Opus 5not measured
A missing bar means that evaluation was not run for that model — it is not a zero.
View as table
| Evaluation | V3.1 Terminus | Claude Opus 5 |
|---|---|---|
| GPQA Diamond | 75.1% | 93.2% |
| Humanity's Last Exam | 8.7% | 54.9% |
| SciCode | 32.1% | 55.7% |
| τ²-bench | 37.1% | 42.1% |
| Terminal-Bench Hard | 31.8% | — |
| LiveCodeBench | 52.9% | — |
| AA-LCR long context | 45.0% | 75.7% |
| AIME 2025 | 53.7% | — |
Estimated from list pricing: 50K input tokens plus 80K output tokens for reasoning models (25K for non-reasoning).