Skip to content
llm-speed
Leaderboard/RTX 3090/Qwen2.5-Coder-14B-Instruct

Can an RTX 3090 run Qwen2.5-Coder-14B-Instruct?

Yes: an RTX 3090 runs Qwen2.5-Coder 14B at 78.2 tok/s at Q4_K_M, measured on llama.cpp with a 131-token prompt in October 2026.

2 signed runs on llm-speed. Single-request decode speed in tokens per second; the short-chat workload unless noted. 1 of 2 runs were measured by llm-speed (@llmspeed); the other one is community-submitted (signed on the submitter's own machine, not measured by llm-speed, and not used for llm-speed verdicts). Runs measured by llm-speed are credited to @llmspeed. Verdicts and headline numbers use only llm-speed's own runs and fit tests.

Measured tok/s by quant and runtime

QuantRuntimeBuildDecode tok/s (best run)WorkloadRunsSource
Q4_K_Mllama.cppcommit ea0c4c278.2chat-short1Measured by llm-speed@llmspeed
Q4_K_Mollama0.31.169.2chat-short1Community-submitted

Fastest quant that fits: Q4_K_M on llama.cpp, 78.2 tok/s (measured by llm-speed; it fits because it ran on this hardware). Quants without a run are not listed.

VRAM: not recorded for these runs. The VRAM calculator gives an estimate (predicted, not measured).

Every measured row

RunDate (UTC)WorkloadQuantRuntime / buildPrompt tokensOutput tokensDecode tok/sPrefill tok/sTTFT msVRAM GiBSource
r_rkyf9wxzw9z2026-10-17chat-shortQ4_K_Mllama.cpp · commit ea0c4c213125678.21693.277–Measured by llm-speed@llmspeed
r_6ahy-dq0f_02026-07-02chat-shortQ4_K_Mollama · 0.31.113125669.2402.0326–Community-submitted
r_rkyf9wxzw9z2026-10-17chat-longQ4_K_Mllama.cpp · commit ea0c4c2316843674.32549.21243–Measured by llm-speed@llmspeed
r_6ahy-dq0f_02026-07-02chat-longQ4_K_Mollama · 0.31.1316853265.82070.61530–Community-submitted
r_rkyf9wxzw9z2026-10-17agent-traceQ4_K_Mllama.cpp · commit ea0c4c2326884075.97427.0273–Measured by llm-speed@llmspeed
r_6ahy-dq0f_02026-07-02agent-traceQ4_K_Mollama · 0.31.1326884063.63518.8483–Community-submitted
r_rkyf9wxzw9z2026-10-17concurrent-decodeQ4_K_Mllama.cpp · commit ea0c4c2102425677.1–––Measured by llm-speed@llmspeed
r_6ahy-dq0f_02026-07-02concurrent-decodeQ4_K_Mollama · 0.31.1102425667.6–––Community-submitted

Questions

Can an RTX 3090 run Qwen2.5-Coder-14B-Instruct?

Yes: an RTX 3090 runs Qwen2.5-Coder 14B at 78.2 tok/s at Q4_K_M, measured on llama.cpp with a 131-token prompt in October 2026.

How many tokens per second does Qwen2.5-Coder-14B-Instruct get on an RTX 3090?

Decode speed measured by llm-speed, by quantization and runtime: Q4_K_M on llama.cpp: 78.2 tok/s. Based on 1 run measured by llm-speed; single-request chat-short unless noted. 1 community-submitted run is listed on the page, labelled, and not counted here.

What is the fastest quant of Qwen2.5-Coder-14B-Instruct that fits on an RTX 3090?

Q4_K_M on llama.cpp was the fastest measured setting at 78.2 tok/s; it fit because it ran. Other quants are listed only where a run measured them.

How much VRAM does Qwen2.5-Coder-14B-Instruct use on an RTX 3090?

Memory use was not recorded for these runs. The VRAM calculator gives an estimate, not a measurement.

Sources and related pages