Model benchmarks
thinkingmachines/inkling-2-mini local LLM performance
As of November 2026, thinkingmachines/inkling-2-mini runs at up to 79.3 tok/s for local inference (best of 4 community benchmark runs across 2 GPUs).
Model size
26B
Peak speed
79.3 tok/s
Average speed
50.7 tok/s
Avg PP
730.5 tok/s
Min memory
11.3 GB
Max context
131,072 tokens
Avg output / run
31,037 tokens
Avg runtime / run
10m 38s
Avg quality
83.2
Benchmark runs
4
GPUs tested
2
Quality by task
Average LLM-judged quality (0–100) with the run-to-run spread shown as a P5–P95 band, overall and for each benchmark task, across all 4 runs. The low and high columns show how much the judge’s score varies between runs, and need at least two runs to display.
| Task | P5 (low) | Avg | P95 (high) |
|---|---|---|---|
| Overall | 79.8 | 83.2 | 87.2 |
| Agent Workflow | 89.6 | 91.2 | 92.7 |
| Code Generation | 81.3 | 83.7 | 86.4 |
| Role Play & Narrative | 56.5 | 71.6 | 87.1 |
| Research & Analysis | 84.2 | 86.4 | 88.3 |
Performance by hardware and tool
Every hardware/tool/quantization combination thinkingmachines/inkling-2-mini has been benchmarked on, ranked by peak token generation speed. Last updated November 2026.
| Hardware | Tool | Quant | Best tok/s | Avg tok/s | Memory | Context | Quality | Runs |
|---|---|---|---|---|---|---|---|---|
| NVIDIA GeForce RTX 3080 Ti | OpenAI-compatible | — | 79.3 tok/s | 79.3 tok/s | 11.3 GB | 65,536 tokens | 79.7 | 1 |
| Apple M5 Max | LM Studio | Q4_K_M | 41.3 tok/s | 41.2 tok/s | 20.0 GB | 131,072 tokens | 84.4 | 3 |
Benchmark runs
All 4 thinkingmachines/inkling-2-mini runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is thinkingmachines/inkling-2-mini good for coding?
- In our benchmarks, thinkingmachines/inkling-2-mini scores 83.7/100 for coding. It runs at about 50.7 tok/s, so if you want more speed, ornith-1.5-35b-a3b-apex-mtp-i-quality is faster (~105.2 tok/s) and still scores well for coding (83.6/100).
- Is thinkingmachines/inkling-2-mini good for agentic (tool-using) tasks?
- In our benchmarks, thinkingmachines/inkling-2-mini scores 91.2/100 for agentic workflows, among the top 3 for agentic workflows on consumer hardware (≤24 GB VRAM). It runs at about 50.7 tok/s, so if you want more speed, Nex-N2.5-mini-APEX-Mini is faster (~88.2 tok/s) and still scores well for agentic workflows (91.3/100).
- How fast is thinkingmachines/inkling-2-mini for local inference?
- Across 4 community benchmark runs, thinkingmachines/inkling-2-mini reaches up to 79.3 tok/s and averages 50.7 tok/s, with the fastest results on NVIDIA GeForce RTX 3080 Ti.
- How much memory does thinkingmachines/inkling-2-mini need?
- The leanest observed configuration used about 11.3 GB of memory (quantizations tested: Q4_K_M).
- Which tools have been used to run thinkingmachines/inkling-2-mini?
- Benchmarks were submitted using LM Studio, OpenAI-compatible. Results are community-contributed and updated as new runs arrive.