Can an RTX 5090 run Qwen3.6-27B?
Yes: an RTX 5090 runs Qwen3.6 27B at 75.1 tok/s at Q4_K_M, measured on llama.cpp with a 112-token prompt in October 2026.
75 signed runs on llm-speed. Single-request decode speed in tokens per second; the short-chat workload unless noted. 1 of 75 runs were measured by llm-speed (@llmspeed); the other 74 are 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
| Quant | Runtime | Build | Decode tok/s (best run) | Workload | Runs | Source |
|---|---|---|---|---|---|---|
| Q4_K_M | llama.cpp | commit 1268eda | 75.1 | chat-short | 1 | Measured by llm-speed@llmspeed |
| Q4_K_M | llama.cpp | not reported | 74.0 | chat-short | 74 | Community-submitted |
Fastest quant that fits: Q4_K_M on llama.cpp, 75.1 tok/s (measured by llm-speed; it fits because it ran on this hardware). Quants without a run are not listed.
Every measured row
| Run | Date (UTC) | Workload | Quant | Runtime / build | Prompt tokens | Output tokens | Decode tok/s | Prefill tok/s | TTFT ms | VRAM GiB | Source |
|---|---|---|---|---|---|---|---|---|---|---|---|
| r_pob-co51bd- | 2026-10-17 | chat-short | Q4_K_M | llama.cpp · commit 1268eda | 112 | 256 | 75.1 | 302.2 | 371 | – | Measured by llm-speed@llmspeed |
| r_f7ulllqu4vk | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 74.0 | – | 190 | – | Community-submitted |
| r_j1cr6asrnn6 | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 73.8 | – | 346 | – | Community-submitted |
| r_az687c4xasr | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 73.6 | – | 167 | – | Community-submitted |
| r_wax_x2ryqhk | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 73.3 | – | 296 | – | Community-submitted |
| r_qe0el0pkw93 | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 73.3 | – | 176 | – | Community-submitted |
| r_qlo6j6lnnh3 | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 73.2 | – | 291 | – | Community-submitted |
| r_7snxv0llk6f | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 72.9 | – | 165 | – | Community-submitted |
| r_p307g-1cdkl | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 72.8 | – | 297 | – | Community-submitted |
| r_uhv-sxt4h97 | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 72.8 | – | 187 | – | Community-submitted |
| r_2g-h0epeovp | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 72.7 | – | 166 | – | Community-submitted |
| r_j0xki0x2kn2 | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 72.6 | – | 170 | – | Community-submitted |
| r_8qwzo79cf0s | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 72.6 | – | 314 | – | Community-submitted |
| r_l-wg5a6o-vg | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 72.6 | – | 172 | – | Community-submitted |
| r_5m7gbxjpw_c | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 72.6 | – | 215 | – | Community-submitted |
| r_7u98v6--uli | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 72.5 | – | 171 | – | Community-submitted |
| r_3un_x00tm0t | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 72.5 | – | 178 | – | Community-submitted |
| r_83zbwf7c17p | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 72.5 | – | 288 | – | Community-submitted |
| r_v2vrkr4uah1 | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 72.5 | – | 172 | – | Community-submitted |
| r_jbit-ru039h | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 72.4 | – | 172 | – | Community-submitted |
| r_p3m8a1lj4kg | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 72.4 | – | 225 | – | Community-submitted |
| r_yluotk909p8 | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 72.4 | – | 167 | – | Community-submitted |
| r_uvy0zplf-sh | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 72.4 | – | 168 | – | Community-submitted |
| r_ercjdbdw2gi | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 72.3 | – | 290 | – | Community-submitted |
| r_4upad7ubcpi | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 72.3 | – | 198 | – | Community-submitted |
| r_432qs45nio9 | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 72.3 | – | 182 | – | Community-submitted |
| r_qlcfyt0mlv6 | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 72.2 | – | 235 | – | Community-submitted |
| r_j1t3p2tae3a | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 72.2 | – | 289 | – | Community-submitted |
| r_g37y03ugmpk | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 72.2 | – | 171 | – | Community-submitted |
| r_fo83sbnxax4 | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 72.0 | – | 317 | – | Community-submitted |
| r_wqhqzea_7xy | 2026-06-30 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 71.9 | – | 530 | – | Community-submitted |
| r_3hda1iz7yd8 | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 71.8 | – | 177 | – | Community-submitted |
| r_gipbd9a8wrn | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 71.5 | – | 213 | – | Community-submitted |
| r_-1k-8qr_v8a | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 71.3 | – | 410 | – | Community-submitted |
| r_mj85rtghwmg | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 71.1 | – | 241 | – | Community-submitted |
| r_carzhcoe-mi | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 71.0 | – | 200 | – | Community-submitted |
| r_u9eqyrsfrgt | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 70.9 | – | 329 | – | Community-submitted |
| r_ea1db-lhv0r | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 70.9 | – | 185 | – | Community-submitted |
| r_wowv__0b5s6 | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 70.4 | – | 269 | – | Community-submitted |
| r_es8a88t0ez0 | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 70.1 | – | 173 | – | Community-submitted |
| r_wmexo75dt0o | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 70.0 | – | 194 | – | Community-submitted |
| r_q5xsw1p-7f8 | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 69.9 | – | 291 | – | Community-submitted |
| r_jrwk2oj1el- | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 69.9 | – | 180 | – | Community-submitted |
| r_bqsunbd6xa8 | 2026-05-02 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 69.9 | – | 3995 | – | Community-submitted |
| r_9ap2a34-le5 | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 69.8 | – | 350 | – | Community-submitted |
| r_bgylub2qqr- | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 69.8 | – | 398 | – | Community-submitted |
| r_5-8ewf89dyr | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 69.8 | – | 343 | – | Community-submitted |
| r_5b-e5ortgcv | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 69.7 | – | 167 | – | Community-submitted |
| r_2673sv7x5m4 | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 69.7 | – | 178 | – | Community-submitted |
| r_mqx-i-thqtq | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 69.6 | – | 308 | – | Community-submitted |
| r_th_ounxmvpx | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 69.6 | – | 177 | – | Community-submitted |
| r_e7_858msfny | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 69.6 | – | 231 | – | Community-submitted |
| r_aj808r0dw53 | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 69.6 | – | 382 | – | Community-submitted |
| r_u4wa_y_y3vt | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 69.6 | – | 227 | – | Community-submitted |
| r_qrpvh5hd9km | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 69.6 | – | 205 | – | Community-submitted |
| r_w482rf72v6z | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 69.6 | – | 174 | – | Community-submitted |
| r_3n3hnvdgwnx | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 69.5 | – | 179 | – | Community-submitted |
| r_qdm3mkehpyx | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 69.5 | – | 188 | – | Community-submitted |
| r_p01v4lc4c11 | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 69.5 | – | 252 | – | Community-submitted |
| r_lblt3fsiqzb | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 69.5 | – | 175 | – | Community-submitted |
| r_4puowxcjcfu | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 69.3 | – | 301 | – | Community-submitted |
| r_q0wcxrit2bz | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 69.2 | – | 342 | – | Community-submitted |
| r_4fxpo_ja16m | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 69.1 | – | 191 | – | Community-submitted |
| r_-cpk_tabogc | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 69.0 | – | 189 | – | Community-submitted |
| r_qsvrisrvp-e | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 68.9 | – | 300 | – | Community-submitted |
| r_xa08qutcmkm | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 68.9 | – | 326 | – | Community-submitted |
| r_v05axrj3dti | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 68.8 | – | 413 | – | Community-submitted |
| r_2o9kb8hy9uk | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 68.6 | – | 183 | – | Community-submitted |
| r_qiq8q_cqfk5 | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 68.0 | – | 241 | – | Community-submitted |
| r_qhp37dyi9n1 | 2026-07-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 67.9 | – | 185 | – | Community-submitted |
| r_1pww-w7p8sd | 2026-05-02 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 67.3 | – | 553 | – | Community-submitted |
| r_79bwm4mq_4l | 2026-04-27 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 66.3 | – | 353 | – | Community-submitted |
| r_kj4fh_mmzj9 | 2026-05-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 47.8 | – | 2833 | – | Community-submitted |
| r_4u7250hj28o | 2026-05-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 45.9 | – | 3089 | – | Community-submitted |
| r__b89kg2iica | 2026-05-01 | chat-short | Q4_K_M | llama.cpp · not reported | 0 | 256 | 39.6 | – | 227 | – | Community-submitted |
| r_pob-co51bd- | 2026-10-17 | chat-long | Q4_K_M | llama.cpp · commit 1268eda | 3182 | 1024 | 74.5 | 2644.8 | 1203 | – | Measured by llm-speed@llmspeed |
| r_pob-co51bd- | 2026-10-17 | agent-trace | Q4_K_M | llama.cpp · commit 1268eda | 3454 | 840 | 74.1 | 4464.5 | 485 | – | Measured by llm-speed@llmspeed |
| r_pob-co51bd- | 2026-10-17 | concurrent-decode | Q4_K_M | llama.cpp · commit 1268eda | 1024 | 256 | 75.1 | – | – | – | Measured by llm-speed@llmspeed |
Questions
Can an RTX 5090 run Qwen3.6-27B?
Yes: an RTX 5090 runs Qwen3.6 27B at 75.1 tok/s at Q4_K_M, measured on llama.cpp with a 112-token prompt in October 2026.
How many tokens per second does Qwen3.6-27B get on an RTX 5090?
Decode speed measured by llm-speed, by quantization and runtime: Q4_K_M on llama.cpp: 75.1 tok/s. Based on 1 run measured by llm-speed; single-request chat-short unless noted. 74 community-submitted runs are listed on the page, labelled, and not counted here.
What is the fastest quant of Qwen3.6-27B that fits on an RTX 5090?
Q4_K_M on llama.cpp was the fastest measured setting at 75.1 tok/s; it fit because it ran. Other quants are listed only where a run measured them.
How much VRAM does Qwen3.6-27B use on an RTX 5090?
Memory use was not recorded for these runs. The VRAM calculator gives an estimate, not a measurement.
Sources and related pages
- Source runs: r_pob-co51bd-, r_f7ulllqu4vk, r_j1cr6asrnn6, r_az687c4xasr, r_wax_x2ryqhk, r_qe0el0pkw93, r_qlo6j6lnnh3, r_7snxv0llk6f, r_p307g-1cdkl, r_uhv-sxt4h97, r_2g-h0epeovp, r_j0xki0x2kn2, r_8qwzo79cf0s, r_l-wg5a6o-vg, r_5m7gbxjpw_c, r_7u98v6--uli, r_3un_x00tm0t, r_83zbwf7c17p, r_v2vrkr4uah1, r_jbit-ru039h, r_p3m8a1lj4kg, r_yluotk909p8, r_uvy0zplf-sh, r_ercjdbdw2gi, r_4upad7ubcpi, r_432qs45nio9, r_qlcfyt0mlv6, r_j1t3p2tae3a, r_g37y03ugmpk, r_fo83sbnxax4, r_wqhqzea_7xy, r_3hda1iz7yd8, r_gipbd9a8wrn, r_-1k-8qr_v8a, r_mj85rtghwmg, r_carzhcoe-mi, r_u9eqyrsfrgt, r_ea1db-lhv0r, r_wowv__0b5s6, r_es8a88t0ez0, r_wmexo75dt0o, r_q5xsw1p-7f8, r_jrwk2oj1el-, r_bqsunbd6xa8, r_9ap2a34-le5, r_bgylub2qqr-, r_5-8ewf89dyr, r_5b-e5ortgcv, r_2673sv7x5m4, r_mqx-i-thqtq, r_th_ounxmvpx, r_e7_858msfny, r_aj808r0dw53, r_u4wa_y_y3vt, r_qrpvh5hd9km, r_w482rf72v6z, r_3n3hnvdgwnx, r_qdm3mkehpyx, r_p01v4lc4c11, r_lblt3fsiqzb, r_4puowxcjcfu, r_q0wcxrit2bz, r_4fxpo_ja16m, r_-cpk_tabogc, r_qsvrisrvp-e, r_xa08qutcmkm, r_v05axrj3dti, r_2o9kb8hy9uk, r_qiq8q_cqfk5, r_qhp37dyi9n1, r_1pww-w7p8sd, r_79bwm4mq_4l, r_kj4fh_mmzj9, r_4u7250hj28o, r__b89kg2iica
- Qwen3.6-27B: every measured GPU and Mac
- RTX 5090: every measured model
- RTX 5090 vs RTX 4090
- RTX 5090 vs RTX 3090
- How runs are measured · open dataset (CC BY 4.0)