toktapethe record, not a recording of it

midagedev

https://github.com/midagedev

72 runs, newest first · showing the first 30

MiMo-V2.6-Flash-MOPD-MXFP4-00001-of-00002.gguf 19.5 tok/s

bloomery0.2.10 (8e55b8ce)mxfp4P30 · G400 · cache 0%ctx 16koffload partialrtx-a6000linux1 stream! 2 caveats

midagedev

2026-10-10

MiMo-V2.6-Flash-MOPD-MXFP4-00001-of-00002.gguf 19.1 tok/sprefill 84.4 tok/s

bloomery0.2.10 (8e55b8ce)mxfp4P5620 · G480 · cache 0%ctx 16koffload partialrtx-a6000linux1 stream

midagedev

2026-10-10

GLM-5.3-Flash-UD-IQ4_XS-00001-of-00005.gguf 25.1 tok/s

bloomery0.2.10 (8e55b8ce)iq3_sP33 · G400 · cache 0%ctx 16koffload partialrtx-a6000linux1 stream! 3 caveats

midagedev

2026-10-10

GLM-5.3-Flash-UD-IQ4_XS-00001-of-00005.gguf 24.6 tok/sprefill 106 tok/s

bloomery0.2.10 (8e55b8ce)iq3_sP5548 · G600 · cache 0%ctx 16koffload partialrtx-a6000linux1 stream! 1 caveat

midagedev

2026-10-10

bloomery 0.2.6 · Qwen3.8 Flash UD-Q4_K_XL · A6000 + 3090 · 3.8K-token prompt, MTP 83.8 tok/sprefill 1450 tok/s

Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf

bloomery0.2.6 (39b13149)q4_KP3847 · G425 · cache 0%ctx 8koffload partialrtx-3090rtx-a6000linux1 stream

midagedev

2026-10-07

One RTX 3090 · Strata 6f32ec0 · Qwen3.8 Flash Next UD-Q4_K_XL 111 GB (same file) · 3 streams (client-timed) 36.7 tok/sprefill 159 tok/s

qwen3.8-flash-next-unsloth-ud-q4_k_xl

openaiUD-Q4_K_XLP201 · G400 · cache 5%rtx-3090rtx-a6000linux3 streams! 4 caveats

midagedev

2026-10-06

One RTX 3090 · Strata 6f32ec0 · Qwen3.8 Flash Next IQ3_XXS 76 GB · 3 streams (client-timed) 69.5 tok/sprefill 292 tok/s

qwen3.8-flash-next-iq3_xxs

openaiIQ3_XXSP202 · G400 · cache 6%rtx-3090rtx-a6000linux3 streams! 4 caveats

midagedev

2026-10-06

One RTX 3090 · bloomery 0.2.1 · Qwen3.8 Flash Next UD-Q4_K_XL 111 GB · 3 streams (client-timed) 53.9 tok/sprefill 325 tok/s

Qwen3.8 Flash Next

openaiP201 · G400 · cache 0%ctx 16128rtx-3090rtx-a6000linux3 streams! 4 caveats

midagedev

2026-10-06

bloomery 0.2.1 · Qwen3.8 Flash Next UD-Q4_K_XL · RTX A6000 · 2 streams 91.3 tok/s

Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf

bloomery0.2.1 (b83f1269)q4_KP33 · G400 · cache 0%ctx 24064offload partialrtx-a6000linux2 streams! 3 caveats

midagedev

2026-10-05

bloomery 0.2.1 · Qwen3.8 Flash Next UD-Q4_K_XL · RTX A6000 · code review, 1500 tokens 91.1 tok/sprefill 732 tok/s

Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf

bloomery0.2.1 (b83f1269)q4_KP538 · G1500 · cache 0%ctx 24064offload partialrtx-a6000linux1 stream! 1 caveat

midagedev

2026-10-05

bloomery 0.2.1 · Qwen3.8 Flash Next UD-Q4_K_XL · RTX A6000 · Korean chat 69.3 tok/s

Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf

bloomery0.2.1 (b83f1269)q4_KP33 · G400 · cache 0%ctx 24064offload partialrtx-a6000linux1 stream! 1 caveat

midagedev

2026-10-05

bloomery 0.2.1 · Qwen3.8 Flash Next UD-Q4_K_XL · RTX A6000 · Rust chat 80.5 tok/s

Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf

bloomery0.2.1 (b83f1269)q4_KP33 · G400 · cache 0%ctx 24064offload partialrtx-a6000linux1 stream! 1 caveat

midagedev

2026-10-05

bloomery 0.2.1 · GLM-5.3 Flash UD-Q4_K_XL · A6000 + 3090 (--place bp) · code review, 2000 tokens 32.4 tok/sprefill 222 tok/s

GLM-5.3-Flash-UD-Q4_K_XL-00001-of-00006.gguf

bloomery0.2.1 (b83f1269)q4_KP497 · G1003 · cache 0%ctx 2koffload partialrtx-3090rtx-a6000linux1 stream! 2 caveats

midagedev

2026-10-05

bloomery 0.2.1 · GLM-5.3 Flash UD-Q4_K_XL · RTX A6000 · 2 streams 34.0 tok/s

GLM-5.3-Flash-UD-Q4_K_XL-00001-of-00006.gguf

bloomery0.2.1 (b83f1269)q4_KP38 · G626 · cache 0% · min 547ctx 2koffload partialrtx-a6000linux2 streams! 4 caveats

midagedev

2026-10-05

bloomery 0.2.1 · GLM-5.3 Flash UD-Q4_K_XL · RTX A6000 · code review, 2000 tokens 30.5 tok/sprefill 199 tok/s

GLM-5.3-Flash-UD-Q4_K_XL-00001-of-00006.gguf

bloomery0.2.1 (b83f1269)q4_KP497 · G1003 · cache 0%ctx 2koffload partialrtx-a6000linux1 stream! 2 caveats

midagedev

2026-10-05

bloomery 0.2.1 · GLM-5.3 Flash UD-Q4_K_XL · RTX A6000 · Korean chat 31.4 tok/s

GLM-5.3-Flash-UD-Q4_K_XL-00001-of-00006.gguf

bloomery0.2.1 (b83f1269)q4_KP42 · G749 · cache 0%ctx 2koffload partialrtx-a6000linux1 stream! 3 caveats

midagedev

2026-10-05

bloomery 0.2.1 · GLM-5.3 Flash UD-Q4_K_XL · RTX A6000 · Rust chat 30.5 tok/s

GLM-5.3-Flash-UD-Q4_K_XL-00001-of-00006.gguf

bloomery0.2.1 (b83f1269)q4_KP33 · G564 · cache 0%ctx 2koffload partialrtx-a6000linux1 stream! 2 caveats

midagedev

2026-10-05

bloomery 0.2.1 · DeepSeek V4.1 Flash Q3_K_M · RTX A6000 · 2 streams 51.3 tok/s

DeepSeek-V4.1-Flash-Q3_K_M-00001-of-00009.gguf

bloomery0.2.1 (b83f1269)q3_KP28 · G400 · cache 0%ctx 32koffload partialrtx-a6000linux2 streams! 1 caveat

midagedev

2026-10-05

bloomery 0.2.1 · DeepSeek V4.1 Flash Q3_K_M · RTX A6000 · code review, 1500 tokens 40.0 tok/sprefill 166 tok/s

DeepSeek-V4.1-Flash-Q3_K_M-00001-of-00009.gguf

bloomery0.2.1 (b83f1269)q3_KP507 · G1500 · cache 0%ctx 32koffload partialrtx-a6000linux1 stream! 1 caveat

midagedev

2026-10-05

bloomery 0.2.1 · DeepSeek V4.1 Flash Q3_K_M · RTX A6000 · Korean chat 47.0 tok/s

DeepSeek-V4.1-Flash-Q3_K_M-00001-of-00009.gguf

bloomery0.2.1 (b83f1269)q3_KP29 · G400 · cache 0%ctx 32koffload partialrtx-a6000linux1 stream! 1 caveat

midagedev

2026-10-05

bloomery 0.2.1 · DeepSeek V4.1 Flash Q3_K_M · RTX A6000 · Rust chat 37.7 tok/s

DeepSeek-V4.1-Flash-Q3_K_M-00001-of-00009.gguf

bloomery0.2.1 (b83f1269)q3_KP26 · G400 · cache 0%ctx 32koffload partialrtx-a6000linux1 stream! 1 caveat

midagedev

2026-10-05

bloomery · GLM-5.3-Flash UD-Q4_K_XL · A6000 alone · Korean chat 42 in / 705 out · residency + NextN MTP draft 30.8 tok/s

GLM-5.3-Flash-UD-Q4_K_XL-00001-of-00006.gguf

bloomery0.1.0 (07ec8f04)q4_KP42 · G705 · cache 0%ctx 4koffload partialrtx-a6000linux1 stream! 3 caveats

midagedev

2026-10-02

bloomery · GLM-5.3-Flash UD-Q4_K_XL · A6000 alone · Rust chat 33 in / 524 out · residency + NextN MTP draft 32.0 tok/s

GLM-5.3-Flash-UD-Q4_K_XL-00001-of-00006.gguf

bloomery0.1.0 (07ec8f04)q4_KP33 · G524 · cache 0%ctx 4koffload partialrtx-a6000linux1 stream! 2 caveats

midagedev

2026-10-02

bloomery · GLM-5.3-Flash UD-Q4_K_XL · A6000 alone · coding review 497 in / 2000 out · residency + NextN MTP draft 31.6 tok/sprefill 195 tok/s

GLM-5.3-Flash-UD-Q4_K_XL-00001-of-00006.gguf

bloomery0.1.0 (07ec8f04)q4_KP497 · G2000 · cache 0%ctx 4koffload partialrtx-a6000linux1 stream! 1 caveat

midagedev

2026-10-02

clef-flash Q4_K_M on bloomery · Korean suite 93.6 ms p50engine 89.0 ms

clef-flash-Q4_K_M.gguf

decision modelbloomeryabff79a5-dirtyQ4_K_Mrtx-3090rtx-a6000linux! 1 caveat

midagedev

2026-10-02

llama.cpp · GLM-5.3-Flash UD-Q4_K_XL · A6000 alone · Korean chat 42 in / 674 out · PR #27754 MTP draft, -ncmoe 38 21.5 tok/s

GLM-5.3-Flash-UD-Q4_K_XL-00001-of-00006.gguf

llama-serverb11036UD-Q4_K_XL321B · 23B activeP42 · G674 · cache 0%ctx 4kfa offoffload partialrtx-a6000linux1 stream! 4 caveats

midagedev

2026-10-02

llama.cpp · GLM-5.3-Flash UD-Q4_K_XL · A6000 alone · Rust chat 33 in / 562 out · PR #27754 MTP draft, -ncmoe 38 22.5 tok/s

GLM-5.3-Flash-UD-Q4_K_XL-00001-of-00006.gguf

llama-serverb11036UD-Q4_K_XL321B · 23B activeP33 · G562 · cache 0%ctx 4kfa offoffload partialrtx-a6000linux1 stream! 3 caveats

midagedev

2026-10-02

llama.cpp · GLM-5.3-Flash UD-Q4_K_XL · A6000 alone · coding review 497 in / 2000 out · PR #27754 MTP draft, -ncmoe 38 20.8 tok/sprefill 104 tok/s

GLM-5.3-Flash-UD-Q4_K_XL-00001-of-00006.gguf

llama-serverb11036UD-Q4_K_XL321B · 23B activeP497 · G2000 · cache 0%ctx 4kfa offoffload partialrtx-a6000linux1 stream! 1 caveat

midagedev

2026-10-02

bloomery · Qwen3.8 Flash Next UD-Q4_K_XL · A6000 alone · coding review 538 in / 1500 out · residency + MTP draft 86.4 tok/sprefill 574 tok/s

Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf

bloomery0.1.0 (b7e9f46b-dirty)q4_KP538 · G1500 · cache 0%ctx 8koffload partialrtx-a6000linux1 stream! 1 caveat

midagedev

2026-10-01

llama.cpp · Qwen3.8 Flash Next UD-Q4_K_XL · A6000 + 3090 · coding review 538 in / 1500 out · -ncmoe 21, no draft 44.3 tok/sprefill 217 tok/s

Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf

llama-serverb11157UD-Q4_K_XL177B · 10B activeP538 · G1500 · cache 0%ctx 8kfa onoffload partialrtx-3090rtx-a6000linux1 stream! 2 caveats

midagedev

2026-10-01

Older runs →

Page views are counted with Cloudflare Web Analytics: no cookies, no fingerprinting. The toktape binary itself sends nothing anywhere.