AMD · CDNA 2
Best AI Models for AMD Instinct MI210 (64 GB)
Runs 70B models well; 120B MoE only just fits.
64 GB HBM2e1638 GB/s181 TFLOPS FP16300 WCDNA 2 2022
| # | Model | Speed | llmrun Score | Quant · VRAM |
|---|---|---|---|---|
| 1 | Qwen3.8 27BAlibabaHow to run → Q4_K_M · 17.9 GB · 56.4 t/s | 56.4 tok/s | Q4_K_M · 17.9 GB | |
| 2 | Qwen3.6 27BAlibaba Q4_K_M · 17.9 GB · 56.4 t/s | 56.4 tok/s | Q4_K_M · 17.9 GB | |
| 3 | Gemma 4 31B ITGoogle Q4_K_M · 21.7 GB · 48.2 t/s | 48.2 tok/s | Q4_K_M · 21.7 GB | |
| 4 | Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB · 114 t/s | 114 tok/s | Q4_K_M · 16.5 GB | |
| 5 | Q4_K_M · 19.3 GB · 84.2 t/s | 84.2 tok/s | Q4_K_M · 19.3 GB |
| # | Model | Speed | Coding Score | Quant · VRAM |
|---|---|---|---|---|
| 1 | Qwen3.8 27BAlibabaHow to run → Q4_K_M · 17.9 GB · 56.4 t/s | 56.4 tok/s | Q4_K_M · 17.9 GB | |
| 2 | Gemma 4 31B ITGoogle Q4_K_M · 21.7 GB · 48.2 t/s | 48.2 tok/s | Q4_K_M · 21.7 GB | |
| 3 | Qwen3.6 35B A3BAlibaba Q4_K_M · 22.0 GB · 91.3 t/s | 91.3 tok/s | Q4_K_M · 22.0 GB | |
| 4 | Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB · 114 t/s | 114 tok/s | Q4_K_M · 16.5 GB | |
| 5 | Q4_K_M · 19.3 GB · 84.2 t/s | 84.2 tok/s | Q4_K_M · 19.3 GB |
| # | Model | Speed | Reasoning Score | Quant · VRAM |
|---|---|---|---|---|
| 1 | Qwen3.8 27BAlibabaHow to run → Q4_K_M · 17.9 GB · 56.4 t/s | 56.4 tok/s | Q4_K_M · 17.9 GB | |
| 2 | Gemma 4 31B ITGoogle Q4_K_M · 21.7 GB · 48.2 t/s | 48.2 tok/s | Q4_K_M · 21.7 GB | |
| 3 | Qwen3.5 27BAlibaba Q4_K_M · 17.9 GB · 56.4 t/s | 56.4 tok/s | Q4_K_M · 17.9 GB | |
| 4 | Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB · 114 t/s | 114 tok/s | Q4_K_M · 16.5 GB | |
| 5 | Q4_K_M · 19.3 GB · 84.2 t/s | 84.2 tok/s | Q4_K_M · 19.3 GB |
| # | Model | Speed | Math Score | Quant · VRAM |
|---|---|---|---|---|
| 1 | Qwen3.8 27BAlibabaHow to run → Q4_K_M · 17.9 GB · 56.4 t/s | 56.4 tok/s | Q4_K_M · 17.9 GB | |
| 2 | Qwen3.6 27BAlibaba Q4_K_M · 17.9 GB · 56.4 t/s | 56.4 tok/s | Q4_K_M · 17.9 GB | |
| 3 | Qwen3.6 35B A3BAlibaba Q4_K_M · 22.0 GB · 91.3 t/s | 91.3 tok/s | Q4_K_M · 22.0 GB | |
| 4 | Gemma 4 31B ITGoogle Q4_K_M · 21.7 GB · 48.2 t/s | 48.2 tok/s | overall | Q4_K_M · 21.7 GB |
| 5 | Q4_K_M · 19.3 GB · 84.2 t/s | 84.2 tok/s | Q4_K_M · 19.3 GB |
| # | Model | Speed ↓ | llmrun Score | Quant · VRAM |
|---|---|---|---|---|
| 1 | Qwen3.5 4BAlibabaHow to run → Q4_K_M · 3.4 GB · 302 t/s | 302 tok/s | Q4_K_M · 3.4 GB | |
| 2 | Qwen3.5 9BAlibaba Q4_K_M · 6.6 GB · 155 t/s | 155 tok/s | Q4_K_M · 6.6 GB | |
| 3 | GPT OSS 20BOpenAI Q4_K_M · 13.1 GB · 135 t/s | 135 tok/s | Q4_K_M · 13.1 GB | |
| 4 | Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB · 114 t/s | 114 tok/s | Q4_K_M · 16.5 GB | |
| 5 | Q4_K_M · 19.3 GB · 84.2 t/s | 84.2 tok/s | Q4_K_M · 19.3 GB |
| # | Model | Speed | Vision Score | Quant · VRAM |
|---|---|---|---|---|
| 1 | Qwen3.8 27BAlibabaHow to run → Q4_K_M · 17.9 GB · 56.4 t/s | 56.4 tok/s | Q4_K_M · 17.9 GB | |
| 2 | Qwen3.6 27BAlibaba Q4_K_M · 17.9 GB · 56.4 t/s | 56.4 tok/s | Q4_K_M · 17.9 GB | |
| 3 | Gemma 4 31B ITGoogle Q4_K_M · 21.7 GB · 48.2 t/s | 48.2 tok/s | Q4_K_M · 21.7 GB | |
| 4 | Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB · 114 t/s | 114 tok/s | Q4_K_M · 16.5 GB | |
| 5 | JEV 27B VLautotrust Q4_K_M · 17.9 GB · 56.4 t/s | 56.4 tok/s | Q4_K_M · 17.9 GB |
See all 373 models on the Instinct MI210 →Best quant that fits · speeds estimated on this card
The Instinct MI210 runs 70B models well. CDNA 2 generation, released 2022.
GenerationCDNA 2released 2022
Instinct MI210 strengths and weaknesses
Strengths
- Runs 120B MoE models: GPT OSS 120B at 96.5 tok/s in 63.5 GB
- Room for long context: Llama 3.3 70B Instruct at Q4 uses 51.3 GB, leaving 12.7 GB free
Weaknesses
- Slow prompt reading: A 4K-token prompt on Qwen3.8 27B waits 5.7 s for the first token
- ROCm, not CUDA: No FP8 weights; MXFP4 only through llama.cpp, not vLLM
Output speed depends on memory bandwidth (1638 GB/s).
- 9BQwen3.5 9B · Q4_K_M · 6.6 GB155tok/s
- 14BQwen3 14B · Q4_K_M · 11.9 GB104tok/s
- 27–31BQwen3.8 27B · Q4_K_M · 17.9 GB56.4tok/s
- 35B MoEQwen3.6 35B A3B · Q4_K_M · 22.0 GB91.3tok/s
- 70BLlama 3.3 70B Instruct · Q4_K_M · 51.3 GB21.1tok/s
- 96.5tok/s
Estimated speeds: Q4_K_M, 4K context, one user. How we estimate
All 1235 models on the Instinct MI210
Featured onlyBest quant that fits, ordered by llmrun Score. Fit grades how well it runs here, S best. How ranking works
| Model | Quant | VRAM | Speed | Context | Score | Fit |
|---|---|---|---|---|---|---|
Tiny Random Qwen3ChatFunctions | Q4_K_M | 0.3 GB | 3276.8 t/s | 33K | S100, Runs great | |
Qwen3 Coder 30B A3B InstructChatCodeFunctions | Q4_K_M | 19.4 GB | 85.5 t/s | 262K | S88, Runs great | |
Qwen3 30B A3B Thinking 2507ChatFunctions | Q4_K_M | 19.4 GB | 85.5 t/s | 262K | S88, Runs great | |
Ornith 1.0 35BChatFunctions | Q4_K_M | 21.5 GB | 95.9 t/s | 262K | S90, Runs great | |
| Q4_K_M | 1.2 GB | 918.7 t/s | 128K | S100, Runs great | ||
LocoOperator 4BChatCodeFunctions | Q4_K_M | 4.2 GB | 339.0 t/s | 262K | S100, Runs great | |
OneRec 1.7BChatFunctions | Q4_K_M | 3.5 GB | 543.1 t/s | 41K | S100, Runs great | |
Llama 3.3 70B InstructChatFunctions | Q4_K_M | 51.3 GB | 21.1 t/s | 131K | B61, Decent | |
Zeta 2.1Chat | Q4_K_M | 7.4 GB | 178.1 t/s | 33K | S99, Runs great | |
DeepSeek OCRVision | Q4_K_M | 2.8 GB | 308.1 t/s | 8K | S100, Runs great | |
Qwen3 0.6BChatFunctions | Q4_K_M | 1.7 GB | 1129.9 t/s | 41K | S100, Runs great | |
| BF16 | 0.0 GB | 98304.0 t/s | 0K | S100, Runs great | ||
GPT OSS 20B RichardErkhov HeresyChatFunctions | Q4_K_M | 13.1 GB | 134.5 t/s | 131K | S95, Runs great | |
| Q4_K_M | 8.9 GB | 121.2 t/s | 131K | S95, Runs great | ||
LLaDA2.0 MiniChatFunctions | Q4_K_M | 10.7 GB | 203.7 t/s | 33K | S100, Runs great |
Specs
- Architecture
- CDNA 2
- VRAM
- 64 GB HBM2e
- Memory bandwidth
- 1638.4 GB/s
- FP16 (dense)
- 181.00 TFLOPS
- Stream processors
- 6,656
- TDP
- 300 W
- Released
- 2022-03-22
Instinct MI210 FAQ
What is the largest model the Instinct MI210 can run?
GPT OSS 120B (116.8B) is the largest at 4-bit or better: at IQ4_XS it uses 63.5 GB of 64 GB, and runs at about 96.5 tok/s.
Can the Instinct MI210 run Qwen3.8 27B?
Yes. Qwen3.8 27B fits at Q4_K_M, using 17.9 GB of 64 GB, and it is the best model that runs well on this card.
How many tokens per second does the Instinct MI210 give?
About 56.4 tok/s on Qwen3.8 27B at Q4_K_M, with the first token after 5.7 s on a 4K-token prompt.
What power supply does the Instinct MI210 need?
Plan a 600 W PSU (2× the 300 W TDP).
Is 64 GB enough for coding and agents?
Yes for coding: Qwen3.8 27B, the best of the 11 coding models that fit, runs at about 56 tok/s. For agents, not yet: no agent-scored model fits in less than 96 GB.