AMD · 8GB VRAM
A Radeon RX 7600 has 8GB of VRAM. Checked against 225 models from the Ollama library, 158 have good/excellent reference fits on the GPU at 4K context and 44 more run with part of the model in system RAM.
158
run well
44
with trade-offs
225
builds evaluated
This is the 8 GB RX 7600, not the 16 GB RX 7600 XT. Ollama lists the RX 7600 for ROCm on both operating systems; the driver/backend must actually be installed.
ollama ps. Intel Macs use the Ollama CPU path; Apple Silicon uses its integrated Metal GPU.Hardware specifications · Ollama support table · reviewed 2026-09-15.
Compare Radeon RX 9060 XT 16GB (16 GB). Use the same model, quantization and context to see what the capacity difference changes.
Editorial examples chosen for their documented tasks, all Q4_K_M at 4,096 tokens. Check the verdict before downloading. These are estimates, not benchmark rankings.
A small instruction-tuned model for trying local chat with more memory headroom. Review summaries against the source text.
Task and build source · Catalog checked 2026-10-13
Est. memory 2.9 GB · medium confidence
Model fully fits in usable GPU VRAM.
ollama run llama3.2:3bAn 8B instruction-tuned alternative to compare on your own prompts. It needs more memory than the 3B example and can still give incorrect answers.
Task and build source · Catalog checked 2026-10-13
Est. memory 5.8 GB · medium confidence
Model fully fits in usable GPU VRAM.
ollama run llama3.1:8bA code-focused instruct build. Review generated changes and run your tests; a memory fit is not evidence that its code is correct.
Task and build source · Catalog checked 2026-10-13
Est. memory 5.3 GB · medium confidence
Model fully fits in usable GPU VRAM.
ollama run qwen2.5-coder:7bA 14B instruct build for comparing answers when memory allows. Larger size does not establish better results for every task.
Task and build source · Catalog checked 2026-10-13
Est. memory 10.2 GB · medium confidence
Estimated 69% of layers offloaded to GPU (33/48); model does not fully fit in VRAM.
ollama run qwen2.5:14bAfter opening Ollama's prompt, enter /set parameter num_ctx 4096 to match this calculation, within the model's supported maximum. Install a current Ollama release that supports the build. Setup and troubleshooting.
Largest first — these have good/excellent reference fits with 8GB of VRAM at 4K context. A good fit can include a small estimated CPU offload. These are technical defaults, not a quality ranking.
| Model | Build | Memory needed | Verdict |
|---|---|---|---|
| nous-hermes2Ollama Library | nous-hermes2:10.7b-solar-q4_K_MQ4_K_M | ~7.6 GB | Good fit |
| solarUpstage | solar:10.7b-text-v1-q4_K_MQ4_K_M | ~7.6 GB | Good fit |
| deepseek-ocrDeepSeek AI | deepseek-ocr:3bF16 | ~7.4 GB | Good fit |
| olmo2Allen Institute for AI | olmo2:7bQ4_K_M | ~6.9 GB | Good fit |
| olmo-3Allen Institute for AI | olmo-3:7b-instruct-q4_K_MQ4_K_M | ~6.9 GB | Good fit |
| codegeex4Zhipu AI | codegeex4:9b-all-q4_K_MQ4_K_M | ~6.8 GB | Good fit |
| glm4Zhipu AI | glm4:9b-text-q4_K_MQ4_K_M | ~6.8 GB | Good fit |
| nimble-v3Ollama Library | nimble-v3:9b-q4_K_MQ4_K_M | ~6.6 GB | Good fit |
| ornith-1.5Ollama Library | ornith-1.5:9bQ4_K_M | ~6.6 GB | Good fit |
| nimbleOllama Library | nimble:9b-q4_K_MQ4_K_M | ~6.6 GB | Good fit |
| ornithOllama Library | ornith:9bQ4_K_M | ~6.6 GB | Good fit |
| nemotron-3-nanoNVIDIA | nemotron-3-nano:4bQ4_K_M | ~6.5 GB | Good fit |
| deepseek-llmDeepSeek AI | deepseek-llm:7b-base-q4_K_MQ4_K_M | ~6.5 GB | Good fit |
| llama2Meta AI | llama2:7b-text-q4_K_MQ4_K_M | ~6.5 GB | Good fit |
| duckdb-nsqlOllama Library | duckdb-nsql:7b-q4_K_MQ4_K_M | ~6.5 GB | Good fit |
| orca2Ollama Library | orca2:7b-q4_K_MQ4_K_M | ~6.5 GB | Good fit |
| wizardcoderOllama Library | wizardcoder:7b-python-q4_K_MQ4_K_M | ~6.5 GB | Good fit |
| wizard-mathOllama Library | wizard-math:7b-q4_K_MQ4_K_MMemory scenario exceeds documented maximum of 2,048 tokens. | ~6.5 GB | Good fit |
| wizardlmOllama Library | wizardlm:7b-q4_K_MQ4_K_MMemory scenario exceeds documented maximum of 2,048 tokens. | ~6.5 GB | Good fit |
| magicoderOllama Library | magicoder:7b-s-cl-q4_K_MQ4_K_M | ~6.5 GB | Good fit |
| llavaOllama Library | llava:7b-v1.5-q4_K_MQ4_K_M | ~6.5 GB | Good fit |
| codellamaMeta AI | codellama:7b-python-q4_K_MQ4_K_M | ~6.5 GB | Good fit |
| meditronOllama Library | meditron:7b-q4_K_MQ4_K_MMemory scenario exceeds documented maximum of 2,048 tokens. | ~6.5 GB | Good fit |
| orca-miniOllama Library | orca-mini:7b-v2-q4_K_MQ4_K_MMemory scenario exceeds documented maximum of 2,048 tokens. | ~6.5 GB | Good fit |
| llama2-chineseMeta AI | llama2-chinese:7b-chat-q4_K_MQ4_K_M | ~6.5 GB | Good fit |
Showing the 25 largest of 158 models that run well.
Each model is evaluated at 4K context against 8GB of VRAM, counting model weights, KV cache, compute buffers and runtime overhead, minus a reserve for the display and operating system. Download sizes come from the official Ollama registry.
Reference scenario: Windows 11, 32GB of system RAM, an illustrative 8-core/16-thread CPU and 500GB free disk. Available RAM, driver version and free VRAM are unknown; the engine applies reserves. Supported GPU drivers/backends are assumed installed. Your own machine may differ — RAM, free disk space and whether an acceleration backend is actually installed all change the answer, which is what the PC scan measures directly.