Reported by the source; GPU count, offloading and concurrent requests can change these figures. Prompt-processing speed is shown only when the source reports it; generation speed cannot tell us time to first token. Check the full setup before comparing.
User reports Qwen3.8-27B at about 124 t/s greedy on a single RTX 3090.
Setup uses fp8 KV cache, int8 lm_head and embed_tokens, MTP-4 drafts, GPTQ-int4 lm_head and MTP module, a split-KV attention kernel, and KVarN 4/2-bit KV cache for 262k context.
Peak concurrent throughput is about 1000 t/s at 64 concurrent.
Reported by the source; GPU count, offloading and concurrent requests can change these figures. Prompt-processing speed is shown only when the source reports it; generation speed cannot tell us time to first token. Check the full setup before comparing.
User reports Qwen3.8 27B at 82 t/s single request on an RTX 3090, with up to 672 t/s peak and 417 t/s sustained at 64 concurrent.
Setup is vLLM with W4A16 quantization, an FP8 KV cache, and int8 lm_head and embed_tokens, at up to 195k context.
The title says Qwen3.8-27b while the body mentions Qwen3.6-28B.
Reported by the source; GPU count, offloading and concurrent requests can change these figures. Prompt-processing speed is shown only when the source reports it; generation speed cannot tell us time to first token. Check the full setup before comparing.
creative-writing
User reports Qwen3.8 27B at about 40 t/s for SVG generation on 2x RTX 3090.
Setup is the cyankiwi/Qwen3.8-27B-AWQ-INT4 model with TP2 and an FP8 KV cache.
Speed started near 60 t/s and dropped to the 20s after the user removed a power limit mid-generation. The run took 20 minutes, 12 of them thinking, and the user was impressed with the result.
Reported by the source; GPU count, offloading and concurrent requests can change these figures. Prompt-processing speed is input throughput, not time to first token. Check the full setup before comparing.
codingagenticvisiontool-use
User reports Qwen3.8-27B at a median 52.2 tok/s decode and 763 tok/s prefill at 111.8K tokens on an Intel Arc Pro B70 32GB.
Setup is vLLM XPU with the GPTQ INT4 quant, MTP2 speculative decoding, an FP8 KV cache, and 128K context.
Vision, tool calling, and agent tests pass. vLLM beats llama.cpp SYCL by about 1.8x. An MTP plus concurrency crash was fixed with max-num-seqs 1, and a prefix caching bug was patched.
Reported by the source; GPU count, offloading and concurrent requests can change these figures. Prompt-processing speed is input throughput, not time to first token. Check the full setup before comparing.
User benchmarks Qwen3.8-27B, identified as Qwen3.5 in llama.cpp, using llama.cpp RPC across multiple GPUs.
Setup is an RX 7900 GRE 16GB as the main host, with RPC combinations including RX 7900 GRE + GTX 1080Ti and GTX 1080Ti + two P102-100, in Q4_K_M and Q6_K quants.
Results vary by config; the best generation is about 12.53 t/s with Q6_K on RX 7900 GRE + GTX 1080Ti.
Reported by the source; GPU count, offloading and concurrent requests can change these figures. Prompt-processing speed is shown only when the source reports it; generation speed cannot tell us time to first token. Check the full setup before comparing.
agenticcoding
User reports that tuning the MTP draft window from n-max 12 to 5 improved throughput from 63 to 99.58 t/s.
The Q4_K_XL quant reached 125.63 t/s, faster than Q8_0 at 99.58 t/s, with no quality gap detected.
The user recommends reasoning_effort medium over xhigh or low, and notes that model quality revealed harness bugs.
Reported by the source; GPU count, offloading and concurrent requests can change these figures. Prompt-processing speed is input throughput, not time to first token. Check the full setup before comparing.
codingvision
User reports Qwen3.8 27B IQ4_XS with vision projector on an M2 MacBook Pro 32GB, at 21.9 t/s prompt and 8.6 t/s generation.
Setup is llama.cpp built from source, with context set to 128K due to memory constraints.
User notes the model is slow on Mac but that thinking quality is good, and mentions previous experience with Qwen3.5 35B-A3B.
Reported by the source; GPU count, offloading and concurrent requests can change these figures. Prompt-processing speed is shown only when the source reports it; generation speed cannot tell us time to first token. Check the full setup before comparing.
codingagentic
User reports Qwen3.8-27B at ~39 tok/s fully GPU-offloaded on an RTX 4080 SUPER 16GB, with ~200-230 ms TTFT on short tasks.
Setup is LM Studio on Windows 11 with the Ubuntu Codex CLI, on a 5800X3D with 3000mhz DDR4.
On a real 26K-token workload the speed is ~34 tok/s. The user finds it surprisingly capable for local coding and agent work but too inconsistent to replace a hosted model, and an excellent local workhorse for bounded tasks.
Reported by the source; GPU count, offloading and concurrent requests can change these figures. Prompt-processing speed is input throughput, not time to first token. Check the full setup before comparing.
User reports 194,048 context tokens at ~1650 prompt t/s and ~40 t/s generation without MTP, and 131,584 context tokens at ~1650 prompt t/s and ~65 t/s generation with MTP, up to 80 t/s for code, on an RTX 4090 FE with 48 GB DDR4 in WSL2.
Setup uses a q8_0 KV cache and q4_0 draft KV cache, with --cache-ram 32768, --ctx-checkpoints 32, --no-mmap and --mlock.
Reported by the source; GPU count, offloading and concurrent requests can change these figures. Prompt-processing speed is shown only when the source reports it; generation speed cannot tell us time to first token. Check the full setup before comparing.
User reports Qwen3.8-27B-heretic-ara Q5_K_M GGUF at ~30-32 t/s on an RTX 3090 with 64 GB DDR4 RAM and an AMD 7950x CPU.
Setup is llama.cpp.
Reported by the source; GPU count, offloading and concurrent requests can change these figures. Prompt-processing speed is shown only when the source reports it; generation speed cannot tell us time to first token. Check the full setup before comparing.
coding
User reports Qwen 3.8 27B at up to 91 t/s with MTP on 3x RTX 3090 plus a Tesla P40 and 128 GB of system RAM.
Setup is the UD-Q8_K_XL quant. Thinking time varies by setting: low about 3s, medium about 3min, xHigh about 15min.
The user compares it to Qwen 3.6 27B and DeepSeek V4 Flash, and says Qwen 3.8 produced a much better Galaga recreation with detailed sprites, sound effects, and a capture system. The user notes the high thinking time may be an issue for slower setups.
Reported by the source; GPU count, offloading and concurrent requests can change these figures. Prompt-processing speed is shown only when the source reports it; generation speed cannot tell us time to first token. Check the full setup before comparing.
codingagentic
User reports Qwen3.8-27B Q8_0 on a Ryzen AI Max+ 395 in a ROG Flow Z13 with 128 GB unified memory, generating at 9-19 t/s and sustaining 16-19 t/s.
Setup is Lemonade Server with llama.cpp ROCm, a Q8 KV cache, native MTP speculative decoding, a 64 GB VRAM and 64 GB RAM split, and roughly 142k context.
MTP acceptance runs 97-99%. The model generated a flight simulator through an agent with file and bash tools.
Reported by the source; GPU count, offloading and concurrent requests can change these figures. Prompt-processing speed is shown only when the source reports it; generation speed cannot tell us time to first token. Check the full setup before comparing.
User compares Qwen 3.8 27B against GLM 5.3 and DeepSeek V4 Flash/Pro, defending it against overthinking criticism.
User mentions 150 t/s decode as a target.
Reported by the source; GPU count, offloading and concurrent requests can change these figures. Prompt-processing speed is input throughput, not time to first token. Check the full setup before comparing.
User reports 7.39 t/s generation and 14 t/s prompt processing on 2x RX 580 8GB, 16 GB VRAM total, on an old DDR3 motherboard with 32 GB RAM.
User calls the setup cheap but low speed.
Reported by the source; GPU count, offloading and concurrent requests can change these figures. Prompt-processing speed is shown only when the source reports it; generation speed cannot tell us time to first token. Check the full setup before comparing.
User reports Qwen3.8 27B Q4_K_M at 25-35 t/s on an RX 7900 XTX at 80k+ context, dropping to 20-21 t/s at 150k-180k context.
Setup is llama.cpp with MTP enabled and a q8_0 KV cache with a q5_1 K cache.
User expresses frustration over the speed and power consumption.
Reported by the source; GPU count, offloading and concurrent requests can change these figures. Prompt-processing speed is shown only when the source reports it; generation speed cannot tell us time to first token. Check the full setup before comparing.
User reports Qwen 3.8 27B at 100+ t/s generation at high context on 4x RTX 3090, each throttled to 200W.
Setup is vLLM with the flashinfer backend, INT8/W8A16, and MTP.
User adds that Gemma models also benefit, but vision models need --language-model-only.
User reports Qwen 3.8 27B overthinking, run on a 4060 Ti with 16 GB VRAM.
The model can be run with 16 GB VRAM plus system RAM. Gemma 4 26B A4B and GPT-OSS 20B are named as alternatives for daily tasks.
The user cites SWE-Rebench insight that Qwen Next requires huge token counts, and hard limits reasoning to 8192 in llama.cpp and vLLM.
Reported by the source; GPU count, offloading and concurrent requests can change these figures. Prompt-processing speed is shown only when the source reports it; generation speed cannot tell us time to first token. Check the full setup before comparing.
User reports Qwen3.8 27B at 19.2 t/s with a Q2_XXS quant and 10.9 t/s with a Q4_KS quant on a 12 GB GPU.
Setup is llama.cpp, with the configuration tuned by GPT-5.6.
Reported by the source; GPU count, offloading and concurrent requests can change these figures. Prompt-processing speed is shown only when the source reports it; generation speed cannot tell us time to first token. Check the full setup before comparing.
User reports Qwen3.8-27B at 50-60 t/s on 4x RTX 5060 Ti 16GB.
Setup is tensor parallelism with MTP.
User asks about memory overclocking risks.
Reported by the source; GPU count, offloading and concurrent requests can change these figures. Prompt-processing speed is shown only when the source reports it; generation speed cannot tell us time to first token. Check the full setup before comparing.
User reports Qwen3.8-27B on an RTX 4090, with throughput ranging from 44 t/s to 134 t/s and no RAM spilling.
Setup includes an mmproj file for vision.