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 | Trinity Large ThinkingArcee AI | Claude Opus 5Anthropic |
|---|---|---|
| Released | Apr 1, 2026 | Jul 24, 2026 |
| Context window | 262K | 1M |
| Max output | 262K | 128K |
| Input / 1M | $0.22 | $5 |
| Output / 1M | $0.85 | $25 |
| Cost per task | $0.08 | $2.25 |
| Arena Elo | 1,159 | 1,393 |
| Serving providers | 2 | 5 |
| Parameters | 398.6B | Undisclosed |
| Licence | other | 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
- Trinity Large Thinking
- Claude Opus 5
Intelligence Index
Trinity Large Thinking18.6Claude Opus 563.1Coding Index
Trinity Large Thinking25.8Claude Opus 578.0Agentic Index
Trinity Large Thinkingnot measuredClaude Opus 559.2Output speed
Trinity Large Thinking67 t/sClaude Opus 587 t/sContext window
Trinity Large Thinking262KClaude Opus 51MLatency · lower is better
Trinity Large Thinking744msClaude Opus 51.12sBlended price / 1M · lower is better
Trinity Large Thinking$0.378Claude Opus 5$10Cost per task · lower is better
Trinity Large Thinking$0.08Claude 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 | Trinity Large Thinking | Claude Opus 5 |
|---|---|---|
| Intelligence Index | 18.6 | 63.1 |
| Coding Index | 25.8 | 78.0 |
| Agentic Index | — | 59.2 |
| Output speed | 67 t/s | 87 t/s |
| Context window | 262K | 1M |
| Latency · lower is better | 744ms | 1.12s |
| Blended price / 1M · lower is better | $0.378 | $10 |
| Cost per task · lower is better | $0.08 | $2.25 |
Evaluation scores
Percentage correct on a common 0–100% scale
- Trinity Large Thinking
- Claude Opus 5
GPQA Diamond
Trinity Large Thinking75.2%Claude Opus 593.2%Humanity's Last Exam
Trinity Large Thinking15.8%Claude Opus 554.9%SciCode
Trinity Large Thinking36.1%Claude Opus 555.7%τ²-bench
Trinity Large Thinking90.1%Claude Opus 542.1%Terminal-Bench Hard
Trinity Large Thinking22.7%Claude Opus 5not measuredAA-LCR long context
Trinity Large Thinking38.3%Claude Opus 575.7%
A missing bar means that evaluation was not run for that model — it is not a zero.
View as table
| Evaluation | Trinity Large Thinking | Claude Opus 5 |
|---|---|---|
| GPQA Diamond | 75.2% | 93.2% |
| Humanity's Last Exam | 15.8% | 54.9% |
| SciCode | 36.1% | 55.7% |
| τ²-bench | 90.1% | 42.1% |
| Terminal-Bench Hard | 22.7% | — |
| LiveCodeBench | — | — |
| AA-LCR long context | 38.3% | 75.7% |
| AIME 2025 | — | — |
Estimated from list pricing: 50K input tokens plus 80K output tokens for reasoning models (25K for non-reasoning).