Agentic
Not comparable- Trinity-Large-Preview
- Not ranked
- Trinity-Large-Thinking
- 41.3
- Estimated · #117/151
- Basis
- BenchAlign lane · 0 vs 1 public rows
- Reading
- Not comparable
Trinity-Large-Preview vs Trinity-Large-Thinking
Updated September 4, 2026. Public scores include evidence status and uncertainty. They are not guarantees for a specific workload. This is a same-family comparison, so migration details appear when the source data supports them.
Decision reading
Trinity-Large-Preview has the higher public score estimate, 55.39 versus 47.12, but the 90% score intervals overlap. Treat that as a lead, not a settled winner.
3 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.
Share or export
Which one for your work
Recommendations appear only when a shared evidence basis or an explicit operating constraint supports the call. Secondary and unsupported use cases stay disclosed below the initial list.
Chat turn cost
1K fresh input + 500 output tokens
Trinity-Large-Thinking
Trinity-Large-Thinking has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.
Confidence: listed-rates
Cache-heavy agent loop cost
200K cached + 20K fresh input + 10K output tokens
Trinity-Large-Thinking
Trinity-Large-Thinking has the lower estimated token cost for this stated workload. Trinity-Large-Preview has no published cached-input rate, so cached tokens use its listed input rate. Trinity-Large-Thinking has no published cached-input rate, so cached tokens use its listed input rate.
Confidence: rate-fallback
Show secondary and unsupported calls
Repository review cost
50K fresh input + 3K output tokens
Trinity-Large-Thinking
Trinity-Large-Thinking has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.
Confidence: listed-rates
Coding work
Code generation, repair, and software-engineering tasks
Not enough matched evidence
Trinity-Large-Preview is not ranked on the public lane for coding, so no winner is named for coding.
Confidence: limited
Agentic work
Tool use, computer use, and multi-step task completion
Not enough matched evidence
Trinity-Large-Preview is not ranked on the public lane for agentic, so no winner is named for agentic.
Confidence: limited
Long documents
Prompts that approach the documented context limit
No clear pick
The documented context windows are equal.
Confidence: documented
What is actually comparable
Shared results can support a head-to-head reading. Results present for only one model describe coverage, not superiority.
1 Trinity-Large-Preview only
3 Shared
2 Trinity-Large-Thinking only
- Shared results
- 3
- Trinity-Large-Preview only
- 1
- Trinity-Large-Thinking only
- 2
- Like-for-like categories
- 0 / 8
Category results, on a stated basis
Each row shows the public-lane category score for both models: the BenchAlign lane for agentic, coding, and knowledge, and the weighted public rows elsewhere, with the same rank each model page reports. A row is like-for-like only when both scores rest on Supported evidence or the same weighted set. Directional and not-comparable rows remain visible, but they never receive a winner.
Coding
Not comparable- Trinity-Large-Preview
- Not ranked
- Trinity-Large-Thinking
- 28.1
- Supported · #172/183
- Basis
- BenchAlign lane · 0 vs 1 public rows
- Reading
- Not comparable
Reasoning
Not comparable- Trinity-Large-Preview
- Not ranked
- Trinity-Large-Thinking
- 48.6
- Unranked · 2 rankable rows
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Knowledge
Not comparable- Trinity-Large-Preview
- Not ranked
- Trinity-Large-Thinking
- 44.1
- Estimated · #121/181
- Basis
- BenchAlign lane · 3 vs 2 public rows
- Reading
- Not comparable
Math
Not comparable- Trinity-Large-Preview
- Not ranked
- Trinity-Large-Thinking
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Multilingual
Not comparable- Trinity-Large-Preview
- Not ranked
- Trinity-Large-Thinking
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Multimodal
Not comparable- Trinity-Large-Preview
- Not ranked
- Trinity-Large-Thinking
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Instruction following
Not comparable- Trinity-Large-Preview
- Not ranked
- Trinity-Large-Thinking
- 67.5
- #64/120
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
| Category | Trinity-Large-Preview | Trinity-Large-Thinking | Basis | Reading |
|---|---|---|---|---|
| Agentic | Not ranked | 41.3Estimated · #117/151 | Not comparableBenchAlign lane · 0 vs 1 public rows | Not comparable |
| Coding | Not ranked | 28.1Supported · #172/183 | Not comparableBenchAlign lane · 0 vs 1 public rows | Not comparable |
| Reasoning | Not ranked | 48.6Unranked · 2 rankable rows | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Knowledge | Not ranked | 44.1Estimated · #121/181 | Not comparableBenchAlign lane · 3 vs 2 public rows | Not comparable |
| Math | Not ranked | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Multilingual | Not ranked | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Multimodal | Not ranked | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Instruction following | Not ranked | 67.5#64/120 | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
Ranks count the models scored in each category’s lane, so the agentic, coding, and knowledge denominators (BenchAlign) differ from the provisional-lane categories. Unranked scores sit on the lane’s scale but never name a winner. Like-for-like rows within 0.5 points read as a practical tie.
Shape of the matched evidence
Only shared public evidence is shown. Sparse evidence stays a ruled list rather than being closed into a radar shape.
A shared-evidence shape is not available.
BenchLM does not draw a radar or infer missing axes when the matched evidence is too sparse.
What each workload costs
Three fixed token mixes turn per-token rates into comparable decisions. Each scenario states context fit and whether cached input had to fall back to the published list-input rate.
Chat turn
1K fresh input + 500 output tokens
- Trinity-Large-Preview
- $0.00075
- Fits in one request
- Trinity-Large-Thinking
- $0.0007
- Fits in one request
Trinity-Large-Thinking has the lower modeled cost
Costs use the listed standard API rates.
Repository review
50K fresh input + 3K output tokens
- Trinity-Large-Preview
- $0.0155
- Fits in one request
- Trinity-Large-Thinking
- $0.0152
- Fits in one request
Trinity-Large-Thinking has the lower modeled cost
Costs use the listed standard API rates.
Cache-heavy agent loop
200K cached + 20K fresh input + 10K output tokens
- Trinity-Large-Preview
- $0.065
- Fits in one request
- Cached input priced at the published list-input rate
- Trinity-Large-Thinking
- $0.064
- Fits in one request
- Cached input priced at the published list-input rate
Trinity-Large-Thinking has the lower modeled cost
Trinity-Large-Preview has no published cached-input rate, so cached tokens use its listed input rate. Trinity-Large-Thinking has no published cached-input rate, so cached tokens use its listed input rate.
Specification differences
Sourced differences are shown directly. Missing facts stay explicit instead of being inferred from a model name or family.
SpecificationTrinity-Large-PreviewTrinity-Large-Thinking
Context window
Maximum documented context; output-token limits may be lower.
Trinity-Large-Preview
512K
Trinity-Large-Thinking
512K
API model ID
Trinity-Large-Preview
Not sourced
Trinity-Large-Thinking
Not sourced
Cached-input rate
A missing cached-input rate falls back to the listed input rate only in the stated workload estimate.
Trinity-Large-Preview
Not published
Trinity-Large-Thinking
Not published
Documented inputs
Trinity-Large-Preview
Not sourced
Trinity-Large-Thinking
Not sourced
Documented outputs
Trinity-Large-Preview
Not sourced
Trinity-Large-Thinking
Not sourced
Provider availability
Trinity-Large-Preview
Not sourced
Trinity-Large-Thinking
Not sourced
Reasoning profile
Trinity-Large-Preview
Non-Reasoning
Trinity-Large-Thinking
Reasoning
Weight access
Trinity-Large-Preview
Open Weight
Trinity-Large-Thinking
Open Weight
License
Trinity-Large-Preview
Open Weight
Trinity-Large-Thinking
Open Weight
Release date
Trinity-Large-Preview
2026-01-27
Trinity-Large-Thinking
2026-03-10
If you are choosing between sibling variants
- Deployment change
- Both entries list Arcee AI as the provider. Confirm endpoint, model ID, limits, and feature support before switching.
- Quality signal
- Trinity-Large-Preview has the higher public score estimate, 55.39 versus 47.12, but the 90% score intervals overlap.
- Workload cost
- Repository review: $0.0155 vs $0.0152. Cache-heavy agent loop: $0.065 vs $0.064.
- Context tradeoff
- Both models list 512K.
Run the same representative tasks against both endpoints before changing production traffic.
Benchmark evidence
The full public result ledger is available for audit without forcing a wide desktop table onto a phone.
Browse raw public benchmark evidence6 rows
Agentic
Gert Labs
Trinity-Large-Preview—Trinity-Large-Thinking32.55%SourceNot directly comparable
Coding
SWE-bench Verified*
Trinity-Large-Preview—Trinity-Large-Thinking63.2%SourceNot directly comparable
Knowledge
MMLU
Trinity-Large-Preview87.2%SourceTrinity-Large-Thinking—Not directly comparable
MMLU-Pro (Arcee)
Trinity-Large-Preview75.2%SourceTrinity-Large-Thinking83.4%SourceTrinity-Large-Thinking leads this result
GPQA-D
Trinity-Large-Preview63.3%SourceTrinity-Large-Thinking76.3%SourceTrinity-Large-Thinking leads this result
Math
Frequently asked questions
Which is better, Trinity-Large-Preview or Trinity-Large-Thinking?
Trinity-Large-Preview has the higher public score estimate, 55.39 versus 47.12, but the 90% score intervals overlap. The higher estimate is not a decisive winner because the uncertainty ranges overlap.
Which is better for coding, Trinity-Large-Preview or Trinity-Large-Thinking?
Trinity-Large-Preview is not ranked on the public lane for coding, so no winner is named for coding.
Which is better for agentic tasks, Trinity-Large-Preview or Trinity-Large-Thinking?
Trinity-Large-Preview is not ranked on the public lane for agentic tasks, so no winner is named for agentic tasks.
Which costs less, Trinity-Large-Preview or Trinity-Large-Thinking?
For the stated presets, chat costs $0.00075 on Trinity-Large-Preview and $0.0007 on Trinity-Large-Thinking; repository review costs $0.0155 and $0.0152; the cache-heavy agent loop costs $0.065 and $0.064. Trinity-Large-Preview has no published cached-input rate, so cached tokens use its listed input rate. Trinity-Large-Thinking has no published cached-input rate, so cached tokens use its listed input rate.
Which has the larger context window, Trinity-Large-Preview or Trinity-Large-Thinking?
Both models list the same context window, 512K.
Related comparisons
Last updated September 4, 2026
Watch Trinity-Large-Preview vs Trinity-Large-Thinking
One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.
Read a sample issueJoin 2,000+ readers.