RX 7900 GRE 16GB

AMD · 16 GB GDDR6 · 576 GB/s · 260 W · MSRP $549 · Search Amazon (affiliate)

The RX 7900 GRE 16GB is a AMD accelerator with 16 GB GDDR6 and about 576 GB/s peak memory bandwidth. AI Hashrate estimates local LLM decode (batch 1) at 8K context: fit counts weights + KV cache + 1 GB runtime overhead — not weights alone. At Q4, about 25 curated models fit fully on this card; at FP16, about 10 fit. Top Q4 speeds: Qwen3-0.6B ≈ 458.2 tok/s (estimated, fits); gemma-3-1b-it ≈ 366.5 tok/s (estimated, fits); Llama-3.2-1B-Instruct ≈ 305.5 tok/s (estimated, fits); Qwen3-1.7B ≈ 183.3 tok/s (estimated, fits); Llama-3.2-3B-Instruct ≈ 114.5 tok/s (estimated, fits). Relative ranking is more reliable than absolute tok/s. See methodology for the bandwidth formula and measured-anchor policy. Methodology.

Table: decode speed estimates (Q4 / FP16). Measured rows override estimates. MSRP is list price, not live retail. Context default 8K.

Model	Params (B)	Quant	tok/s	Fits?
Qwen3-0.6B	0.8	Q4	458.2 est.	Yes
gemma-3-1b-it	1.0	Q4	366.5 est.	Yes
Llama-3.2-1B-Instruct	1.2	Q4	305.5 est.	Yes
Qwen3-30B-A3B	30.5	Q4	264.5 est.	No
Qwen3-Coder-30B-A3B-Instruct	30.5	Q4	264.5 est.	No
Qwen3-30B-A3B-Instruct-2507	30.5	Q4	264.5 est.	No
GLM-4.7-Flash	31.2	Q4	253.9 est.	No
Qwen3-1.7B	2.0	Q4	183.3 est.	Yes
Qwen3-0.6B	0.8	FP16	126.0 est.	Yes
Llama-3.2-3B-Instruct	3.2	Q4	114.5 est.	Yes
gpt-oss-20b	21.5	Q4	101.8 est.	Yes
gemma-3-1b-it	1.0	FP16	100.8 est.	Yes
Phi-4-mini-instruct	3.8	Q4	96.5 est.	Yes
Phi-4-mini-reasoning	3.8	Q4	96.5 est.	Yes
gemma-3-4b-it	4.3	Q4	94.0 est.	Yes
Qwen3-4B-Instruct-2507	4.0	Q4	91.6 est.	Yes
Qwen3-4B-Thinking-2507	4.0	Q4	91.6 est.	Yes
Llama-3.2-1B-Instruct	1.2	FP16	84.0 est.	Yes
Mistral-7B-Instruct-v0.3	7.2	Q4	50.9 est.	Yes
Qwen3-1.7B	2.0	FP16	50.4 est.	Yes
Qwen2.5-7B-Instruct	7.6	Q4	48.2 est.	Yes
Qwen2.5-Coder-7B-Instruct	7.6	Q4	48.2 est.	Yes
DeepSeek-R1-Distill-Qwen-7B	7.6	Q4	48.2 est.	Yes
Llama-3.1-8B-Instruct	8.0	Q4	45.8 est.	Yes
Qwen3-8B	8.2	Q4	44.7 est.	Yes
DeepSeek-R1-0528-Qwen3-8B	8.2	Q4	44.7 est.	Yes
Qwen3-Coder-Next	79.7	Q4	43.3 est.	No
Qwen3-Next-80B-A3B-Instruct	81.3	Q4	41.7 est.	No
Mistral-7B-Instruct-v0.3	7.2	FP16	36.1 est.	No
Devstral-Small-2507	23.6	Q4	32.5 est.	No
Llama-3.2-3B-Instruct	3.2	FP16	31.5 est.	Yes
Qwen2.5-7B-Instruct	7.6	FP16	30.9 est.	No
Qwen2.5-Coder-7B-Instruct	7.6	FP16	30.9 est.	No
DeepSeek-R1-Distill-Qwen-7B	7.6	FP16	30.9 est.	No
Mistral-Small-3.2-24B-Instruct-2506	24.0	Q4	30.9 est.	No
Magistral-Small-2509	24.0	Q4	30.9 est.	No
gemma-3-12b-it	12.2	Q4	30.5 est.	Yes
Llama-3.1-8B-Instruct	8.0	FP16	26.6 est.	No
Phi-4-mini-instruct	3.8	FP16	26.5 est.	Yes
Phi-4-mini-reasoning	3.8	FP16	26.5 est.	Yes
gemma-3-4b-it	4.3	FP16	25.8 est.	Yes
Qwen3-4B-Instruct-2507	4.0	FP16	25.2 est.	Yes
Qwen3-4B-Thinking-2507	4.0	FP16	25.2 est.	Yes
Phi-4	14.7	Q4	24.9 est.	Yes
Phi-4-reasoning	14.7	Q4	24.9 est.	Yes
Qwen3-8B	8.2	FP16	24.8 est.	No
Qwen3-14B	14.8	Q4	24.8 est.	Yes
Qwen2.5-14B-Instruct	14.8	Q4	24.8 est.	Yes
DeepSeek-R1-0528-Qwen3-8B	8.2	FP16	24.8 est.	No
Qwen2.5-Coder-14B-Instruct	14.8	Q4	24.8 est.	Yes
DeepSeek-R1-Distill-Qwen-14B	14.8	Q4	24.8 est.	Yes
Mixtral-8x7B-Instruct-v0.1	46.7	Q4	24.3 est.	No
gemma-3-27b-it	27.4	Q4	21.3 est.	No
Qwen3-32B	32.8	Q4	12.5 est.	No
Qwen2.5-32B-Instruct	32.8	Q4	12.5 est.	No
Qwen2.5-Coder-32B-Instruct	32.8	Q4	12.5 est.	No
DeepSeek-R1-Distill-Qwen-32B	32.8	Q4	12.5 est.	No
gpt-oss-20b	21.5	FP16	10.2 est.	No
gemma-3-12b-it	12.2	FP16	7.9 est.	No
Qwen3-30B-A3B	30.5	FP16	6.3 est.	No
Qwen3-Coder-30B-A3B-Instruct	30.5	FP16	6.3 est.	No
Qwen3-30B-A3B-Instruct-2507	30.5	FP16	6.3 est.	No
GLM-4.7-Flash	31.2	FP16	6.0 est.	No
Phi-4	14.7	FP16	4.5 est.	No
Phi-4-reasoning	14.7	FP16	4.5 est.	No
Qwen3-14B	14.8	FP16	4.4 est.	No
Qwen2.5-14B-Instruct	14.8	FP16	4.4 est.	No
Qwen2.5-Coder-14B-Instruct	14.8	FP16	4.4 est.	No
DeepSeek-R1-Distill-Qwen-14B	14.8	FP16	4.4 est.	No
Llama-3.1-70B-Instruct	70.6	Q4	1.3 est.	No
Llama-3.3-70B-Instruct	70.6	Q4	1.3 est.	No
DeepSeek-R1-Distill-Llama-70B	70.6	Q4	1.3 est.	No
Qwen2.5-72B-Instruct	72.7	Q4	1.2 est.	No
Mistral-Small-3.2-24B-Instruct-2506	24.0	FP16	1.1 est.	No
Magistral-Small-2509	24.0	FP16	1.1 est.	No
Devstral-Small-2507	23.6	FP16	1.1 est.	No
gemma-3-27b-it	27.4	FP16	0.7 est.	No

Fit checks

• Qwen3-0.6B — fits · ≈458.2 tok/s Q4 est.
• Qwen3-8B — fits · ≈44.7 tok/s Q4 est.
• Qwen2.5-7B-Instruct — fits · ≈48.2 tok/s Q4 est.
• Llama-3.2-1B-Instruct — fits · ≈305.5 tok/s Q4 est.
• Llama-3.1-8B-Instruct — fits · ≈45.8 tok/s Q4 est.
• gpt-oss-20b — fits · ≈101.8 tok/s Q4 est.
• Qwen3-32B — does not fit · ≈12.5 tok/s Q4 est.
• Qwen3-1.7B — fits · ≈183.3 tok/s Q4 est.
• Mistral-7B-Instruct-v0.3 — fits · ≈50.9 tok/s Q4 est.
• Qwen3-14B — fits · ≈24.8 tok/s Q4 est.
• Qwen3-4B-Instruct-2507 — fits · ≈91.6 tok/s Q4 est.

All 61 model fit checks — see the linked Fits? column in the table above