Kimi K3 vs Qwen3.8 Max
Qwen3.8 Max and Kimi K3 score 79.3% vs 56.0% averaged across five Vision Evals tasks (#9 and #45 of 65 models), so Qwen3.8 Max leads overall.
- Qwen3.8 Max wins all 5 tasks, with its widest lead on counting (81.1% vs 46.0%).
- Kimi K3 is about half the cost: $0.0049 vs $0.0080 overall average cost per sample.
- Kimi K3 is faster: 9.9s vs 18.9s per sample.
Roboflow Vision Evals. Full breakdown below.
Compare Kimi K3 vs Qwen3.8 Max live
Run the same image across every model that supports a task and compare their outputs side-by-side.
Detect and compare bounding boxes across models on the same image.
Upload an image
Drag and drop an image here, or click to browse
Kimi K3 vs Qwen3.8 Max on Vision Evals
Roboflow Vision Evals. Full breakdown below.
Kimi K3 vs Qwen3.8 Max Comparison Table
Evals updated October 26, 2026Pricing updated November 3, 2026
| Property | Kimi K3 | Qwen3.8 Max |
|---|---|---|
| Organization | Moonshot AI | Qwen |
| Category | open | closed |
| Modality | multimodal | multimodal |
| Release Date | Jul 2026 | Aug 2026 |
| Context Window | 1.0M | 984K |
| Parameters | 2.8T | 2.4T total, ~95B active |
| License | Modified MIT | Apache 2.0 |
| Commercial use | See license terms | free for commercial use |
| Pricing per 1M tokens | ||
| Input $/1M | $0.192 | No published price |
| Output $/1M | $15.00 | No published price |
| Vision Tasks | ||
| Model Features | ||
Both support 3 more features | ||
Vision Evalsground-truth scores across 5 vision tasks, pooled at low effort | ||
| Overall | 56.0% | 79.3% |
| Avg cost / sample | $0.0049 | $0.0080 |
| Avg speed / sample | 9.86s | 18.95s |
| By task | ||
| Object Detection (low) | 51.9% | 76.7% ±0.3, Mean of 3 runs, range 76.5 to 77.1 |
| Object Detection (high) | – | 78.4% ±0.4, Mean of 3 runs, range 78.1 to 78.9 |
| Counting (low) | 46.0% | 81.1% ±2.0, Mean of 3 runs, range 78.4 to 82.4 |
| Counting (high) | – | 81.1% ±0.0, Mean of 3 runs, range 81.1 to 81.1 |
| Identification (low) | 81.3% | 88.5% ±1.6, Mean of 3 runs, range 87.5 to 90.6 |
| Identification (high) | – | 89.6% ±1.6, Mean of 3 runs, range 87.5 to 90.6 |
| OCR (low) | 58.6% | 74.0% |
| by category |
|
|
| OCR (high) | 61.8% | 71.8% |
| by category |
|
|
| Reasoning (low) | 42.4% | 75.9% ±2.0, Mean of 3 runs, range 73.5 to 77.5 |
| Reasoning (high) | 74.2% | 80.3% ±2.0, Mean of 3 runs, range 78.2 to 82.1 |
Kimi K3 vs Qwen3.8 Max: Overview
Kimi K3 is a sparse Mixture-of-Experts large language model developed by Moonshot AI, with 2.8 trillion total parameters and a 1-million-token context window. The model activates 16 out of 896 experts per token using the Stable LatentMoE framework, and is built on two architectural innovations: Kimi Delta Attention (KDA), a hybrid linear attention mechanism that enables up to 6.3x faster decoding in long-context settings, and Attention Residuals (AttnRes), which selectively retrieves representations across model depth and delivers roughly 25% higher training efficiency. Together with refined training and data recipes, these structural advances yield approximately 2.5x better overall scaling efficiency compared to its predecessor Kimi K2. The model applies quantization-aware training from the supervised fine-tuning stage onward, using MXFP4 weights with MXFP8 activations for hardware compatibility. Thinking mode is always enabled at launch, with reasoning effort configurable via the reasoning_effort field.
Kimi K3 supports native visual understanding alongside text, accepting image inputs for tasks that combine software engineering and visual reasoning. It targets long-horizon coding, knowledge work, and agentic workflows, and ships in two variants: K3 Max for general chat and agent tasks, and K3 Swarm Max for large-scale parallel processing across many coordinated sub-agents. The model is compatible with the OpenAI SDK via an OpenAI-compatible API. Full model weights are scheduled for release by July 27, 2026 under a Modified MIT license, following the open-weight pattern established by the Kimi K2 model family. A technical report with full architecture, training, and evaluation details is expected to accompany the weights release.
Qwen3.8 Max is the flagship tier of Alibaba's Qwen3.8 family, a sparse mixture-of-experts multimodal model with roughly 2.4 trillion total parameters of which about 95 billion activate per token, which keeps serving cost and latency well below what the total parameter count would imply. It builds on the architectural foundation established by Qwen3.5 and accepts text, images, video, and documents as input while producing text output. Reported context handling reaches close to one million tokens, with a maximum generation length of 131,072 tokens, so the model is aimed at long-horizon agentic work such as repository-scale coding, multi-step research, data analysis, and office document workflows.
For vision work the model performs image and video understanding, document and chart interpretation, text recognition inside images, and grounded visual question answering, and Alibaba reports gains concentrated in multimodal and agentic evaluation categories rather than general reasoning. It is the first model in the Max tier of the Qwen line for which the team states weights will be released publicly, alongside a smaller Qwen3.8 27B checkpoint. No training or safety model card has been published.