Agentic
Directional only- Mellum2-12B-A2.5B-Thinking
- 48.3
- Estimated · #76/151
- Nemotron 3 Ultra
- 22.2
- Supported · #150/151
- Basis
- BenchAlign lane · 1 vs 6 public rows
- Reading
- Directional only
Mellum2-12B-A2.5B-Thinking vs Nemotron 3 Ultra
Updated September 4, 2026. Public scores include evidence status and uncertainty. They are not guarantees for a specific workload.
Decision reading
At least one model is not scored in the current public ranking lane, so the page does not name an overall quality 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.
Long documents
Prompts that approach the documented context limit
Nemotron 3 Ultra
Nemotron 3 Ultra has the larger documented context window.
Confidence: documented
Show secondary and unsupported calls
Coding work
Code generation, repair, and software-engineering tasks
Not enough matched evidence
Mellum2-12B-A2.5B-Thinking 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
Mellum2-12B-A2.5B-Thinking is scored on Estimated evidence for agentic, so the reading is directional rather than like-for-like.
Confidence: limited
Chat turn cost
1K fresh input + 500 output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
Confidence: listed-rates
Cache-heavy agent loop cost
200K cached + 20K fresh input + 10K output tokens
Not enough matched evidence
The page does not recommend a cost winner because at least one model cannot fit the stated workload in one request. Mellum2-12B-A2.5B-Thinking does not fit this workload in one request. Mellum2-12B-A2.5B-Thinking has no comparable published API token rate. Nemotron 3 Ultra has no comparable published API token rate.
Confidence: listed-rates
Repository review cost
50K fresh input + 3K output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
Confidence: listed-rates
What is actually comparable
Shared results can support a head-to-head reading. Results present for only one model describe coverage, not superiority.
3 Mellum2-12B-A2.5B-Thinking only
3 Shared
21 Nemotron 3 Ultra only
- Shared results
- 3
- Mellum2-12B-A2.5B-Thinking only
- 3
- Nemotron 3 Ultra only
- 21
- Like-for-like categories
- 0 / 8
2 categories rest on Estimated evidence or different evidence sets. Those rows remain visible for coverage context but do not name a winner.
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.
Knowledge
Directional only- Mellum2-12B-A2.5B-Thinking
- 48.8
- Estimated · #96/181
- Nemotron 3 Ultra
- 45.6
- Estimated · #114/181
- Basis
- BenchAlign lane · 3 vs 7 public rows
- Reading
- Directional only
Coding
Not comparable- Mellum2-12B-A2.5B-Thinking
- Not ranked
- Nemotron 3 Ultra
- 26.9
- Supported · #174/183
- Basis
- BenchAlign lane · 1 vs 7 public rows
- Reading
- Not comparable
Reasoning
Not comparable- Mellum2-12B-A2.5B-Thinking
- Not ranked
- Nemotron 3 Ultra
- 48.7
- Unranked · 1 rankable row
- Basis
- Provisional lane · 0 vs 1 weighted rows
- Reading
- Not comparable
Math
Not comparable- Mellum2-12B-A2.5B-Thinking
- Not ranked
- Nemotron 3 Ultra
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Multilingual
Not comparable- Mellum2-12B-A2.5B-Thinking
- Not ranked
- Nemotron 3 Ultra
- 47.4
- #7/12
- Basis
- Provisional lane · 0 vs 1 weighted rows
- Reading
- Not comparable
Multimodal
Not comparable- Mellum2-12B-A2.5B-Thinking
- Not ranked
- Nemotron 3 Ultra
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Instruction following
Not comparable- Mellum2-12B-A2.5B-Thinking
- Not ranked
- Nemotron 3 Ultra
- 88.7
- #28/120
- Basis
- Provisional lane · 0 vs 1 weighted rows
- Reading
- Not comparable
| Category | Mellum2-12B-A2.5B-Thinking | Nemotron 3 Ultra | Basis | Reading |
|---|---|---|---|---|
| Agentic | 48.3Estimated · #76/151 | 22.2Supported · #150/151 | Directional onlyBenchAlign lane · 1 vs 6 public rows | Directional only |
| Knowledge | 48.8Estimated · #96/181 | 45.6Estimated · #114/181 | Directional onlyBenchAlign lane · 3 vs 7 public rows | Directional only |
| Coding | Not ranked | 26.9Supported · #174/183 | Not comparableBenchAlign lane · 1 vs 7 public rows | Not comparable |
| Reasoning | Not ranked | 48.7Unranked · 1 rankable row | Not comparableProvisional lane · 0 vs 1 weighted rows | Not comparable |
| Math | Not ranked | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Multilingual | Not ranked | 47.4#7/12 | Not comparableProvisional lane · 0 vs 1 weighted rows | Not comparable |
| Multimodal | Not ranked | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Instruction following | Not ranked | 88.7#28/120 | Not comparableProvisional lane · 0 vs 1 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.
Too few matched category axes support a radar. The ruled list below shows only shared benchmark results; positions use each benchmark’s normalized display scale when available.
GPQA
Knowledge
Mellum2-12B-A2.5B-Thinking: 57.6%Nemotron 3 Ultra: 87%Normalized gap 29.4Mellum2-12B-A2.5B-Thinking source Nemotron 3 Ultra source
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
- Mellum2-12B-A2.5B-Thinking
- API rate not published
- Fits in one request
- Nemotron 3 Ultra
- Self-hosted; infrastructure cost varies
- Fits in one request
Mellum2-12B-A2.5B-Thinking has no comparable published API token rate. Nemotron 3 Ultra has no comparable published API token rate.
Repository review
50K fresh input + 3K output tokens
- Mellum2-12B-A2.5B-Thinking
- API rate not published
- Fits in one request
- Nemotron 3 Ultra
- Self-hosted; infrastructure cost varies
- Fits in one request
Mellum2-12B-A2.5B-Thinking has no comparable published API token rate. Nemotron 3 Ultra has no comparable published API token rate.
Cache-heavy agent loop
200K cached + 20K fresh input + 10K output tokens
- Mellum2-12B-A2.5B-Thinking
- API rate not published
- Does not fit in one request
- Cached-input rate unavailable
- Nemotron 3 Ultra
- Self-hosted; infrastructure cost varies
- Fits in one request
- Cached-input rate unavailable
Mellum2-12B-A2.5B-Thinking does not fit this workload in one request. Mellum2-12B-A2.5B-Thinking has no comparable published API token rate. Nemotron 3 Ultra has no comparable published API token rate.
Specification differences
Sourced differences are shown directly. Missing facts stay explicit instead of being inferred from a model name or family.
SpecificationMellum2-12B-A2.5B-ThinkingNemotron 3 Ultra
Context window
Maximum documented context; output-token limits may be lower.
Mellum2-12B-A2.5B-Thinking
128K
Nemotron 3 Ultra
1M
API model ID
Mellum2-12B-A2.5B-Thinking
Not sourced
Nemotron 3 Ultra
Not sourced
Cached-input rate
A missing cached-input rate falls back to the listed input rate only in the stated workload estimate.
Mellum2-12B-A2.5B-Thinking
No comparable hosted API rate
Nemotron 3 Ultra
No comparable hosted API rate
Documented inputs
Mellum2-12B-A2.5B-Thinking
Not sourced
Nemotron 3 Ultra
Not sourced
Documented outputs
Mellum2-12B-A2.5B-Thinking
Not sourced
Nemotron 3 Ultra
Not sourced
Provider availability
Mellum2-12B-A2.5B-Thinking
Not sourced
Nemotron 3 Ultra
Not sourced
Reasoning profile
Mellum2-12B-A2.5B-Thinking
Reasoning
Nemotron 3 Ultra
Reasoning
Weight access
Mellum2-12B-A2.5B-Thinking
Open Weight
Nemotron 3 Ultra
Open Weight
License
Mellum2-12B-A2.5B-Thinking
Open Weight
Nemotron 3 Ultra
Open Weight
Release date
Mellum2-12B-A2.5B-Thinking
2026-05-28
Nemotron 3 Ultra
2026-06-04
If you already use one of these models
- Deployment change
- The models list different providers, so authentication, endpoint behavior, limits, and feature support may change.
- Quality signal
- At least one model is not scored in the current public ranking lane, so the page does not name an overall quality winner.
- Workload cost
- A complete comparable API-rate estimate is not available for both models.
- Context tradeoff
- Nemotron 3 Ultra has the larger documented window (1M).
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 evidence27 rows
Agentic
BFCL v4
Mellum2-12B-A2.5B-Thinking45.6%SourceNemotron 3 Ultra—Not directly comparable
Terminal-Bench 2.0
Mellum2-12B-A2.5B-Thinking—Nemotron 3 Ultra56.4%SourceNot directly comparable
PinchBench
Mellum2-12B-A2.5B-Thinking—Nemotron 3 Ultra90.0%SourceNot directly comparable
BrowseComp
Mellum2-12B-A2.5B-Thinking—Nemotron 3 Ultra44.4%SourceNot directly comparable
τ³-bench results
Mellum2-12B-A2.5B-Thinking—Nemotron 3 Ultra70.9%SourceNot directly comparable
HLE w/ tools
Mellum2-12B-A2.5B-Thinking—Nemotron 3 Ultra37.4%SourceNot directly comparable
Terminal-Bench 2.1 (Vals)
Mellum2-12B-A2.5B-Thinking—Nemotron 3 Ultra50.9%SourceNot directly comparable
Coding
LiveCodeBench v6
Mellum2-12B-A2.5B-Thinking69.9%SourceNemotron 3 Ultra89.0%SourceNemotron 3 Ultra leads this result
SWE-bench Verified
Mellum2-12B-A2.5B-Thinking—Nemotron 3 Ultra71.9%SourceNot directly comparable
SWE Multilingual
Mellum2-12B-A2.5B-Thinking—Nemotron 3 Ultra67.7%SourceNot directly comparable
SciCode
Mellum2-12B-A2.5B-Thinking—Nemotron 3 Ultra44.6%SourceNot directly comparable
Terminal-Bench 2.0
Mellum2-12B-A2.5B-Thinking—Nemotron 3 Ultra56.4%SourceNot directly comparable
LiveCodeBench (Vals)
Mellum2-12B-A2.5B-Thinking—Nemotron 3 Ultra86.0%SourceNot directly comparable
SWE-bench (Vals)
Mellum2-12B-A2.5B-Thinking—Nemotron 3 Ultra69.0%SourceNot directly comparable
Reasoning
Knowledge
MMLU-Redux
Mellum2-12B-A2.5B-Thinking86.2%SourceNemotron 3 Ultra—Not directly comparable
GPQA
Mellum2-12B-A2.5B-Thinking57.6%SourceNemotron 3 Ultra87%SourceNemotron 3 Ultra leads this result
GPQA-D
Mellum2-12B-A2.5B-Thinking57.6%SourceNemotron 3 Ultra87.0%SourceNemotron 3 Ultra leads this result
HLE
Mellum2-12B-A2.5B-Thinking—Nemotron 3 Ultra26.7%SourceNot directly comparable
HLE w/o tools
Mellum2-12B-A2.5B-Thinking—Nemotron 3 Ultra26.7%SourceNot directly comparable
MMLU-Pro
Mellum2-12B-A2.5B-Thinking—Nemotron 3 Ultra86.8%SourceNot directly comparable
GPQA Diamond (Vals)
Mellum2-12B-A2.5B-Thinking—Nemotron 3 Ultra86.1%SourceNot directly comparable
MMLU-Pro (Vals)
Mellum2-12B-A2.5B-Thinking—Nemotron 3 Ultra85.8%SourceNot directly comparable
Multilingual
MMLU-ProX
Mellum2-12B-A2.5B-Thinking—Nemotron 3 Ultra83%SourceNot directly comparable
Instruction following
Frequently asked questions
Which is better, Mellum2-12B-A2.5B-Thinking or Nemotron 3 Ultra?
At least one model is not scored in the current public ranking lane, so the page does not name an overall quality winner. The page therefore keeps the decision tied to the specific documented workload.
Which is better for coding, Mellum2-12B-A2.5B-Thinking or Nemotron 3 Ultra?
Mellum2-12B-A2.5B-Thinking is not ranked on the public lane for coding, so no winner is named for coding.
Which is better for agentic tasks, Mellum2-12B-A2.5B-Thinking or Nemotron 3 Ultra?
Mellum2-12B-A2.5B-Thinking scores higher for agentic tasks on the public lane, 48.3 to 22.2. Mellum2-12B-A2.5B-Thinking is scored on Estimated evidence for agentic tasks, so the reading is directional rather than like-for-like. BenchLM does not name a winner for agentic tasks from a directional reading; read the shared benchmark rows directly and test the models on the same task set.
Which costs less, Mellum2-12B-A2.5B-Thinking or Nemotron 3 Ultra?
Both models do not have comparable published API token rates, so this page does not name a universal price winner.
Which has the larger context window, Mellum2-12B-A2.5B-Thinking or Nemotron 3 Ultra?
Nemotron 3 Ultra has the larger documented context window: 1M, compared with 128K.
Related comparisons
Last updated September 4, 2026
Watch Mellum2-12B-A2.5B-Thinking vs Nemotron 3 Ultra
One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.
Read a sample issueJoin 2,000+ readers.