Four months is a long time in open-weight AI. Since this reference first went up, Google moved Gemma to Apache 2.0, DeepSeek shipped V4, Alibaba shipped Qwen 3.6, and Moonshot put a 2.8 trillion parameter model on Hugging Face. This revision folds all of them in, along with OpenAI’s gpt-oss pair, the first OpenAI open weights since GPT-2. The box you choose still decides which of these fit and how fast they generate, so it helps to compare a Mac mini, a mini PC, and a GPU for local LLMs and to understand why unified memory and VRAM behave so differently.
This reference covers every major open-source and open-weight large language model, with verified benchmark scores, license terms, hardware requirements, and hands-on performance data from real self-hosting tests. The tables below compare architecture details, benchmark results, licensing restrictions, and what each model actually needs to run on your own machine using Ollama or llama.cpp.
Every spec here was re-verified against official model cards and release announcements in August 2026. The hands-on numbers come from our own lab runs: the small-model CPU tests from March 2026 on Ubuntu 24.04 (4 vCPUs, 16 GB RAM, Ollama), and the DeepSeek V4 Flash and Kimi K3 measurements from August 2026.
Master Comparison Table
This table covers every major open-source/open-weight LLM family. “Active params” refers to how many parameters are used per inference pass in Mixture-of-Experts (MoE) architectures. Dense models use all parameters on every token.
| Model | Developer | Total Params | Active Params | Architecture | Context Window | Multimodal | License | Release |
|---|---|---|---|---|---|---|---|---|
| Kimi K3 | Moonshot AI | 2.8T | 104B | MoE (896 experts) | 1M tokens | Text + Image + Video | Kimi K3 (custom) | Jul 2026 |
| Kimi K2 | Moonshot AI | 1T | 32B | MoE (384 experts) | 128K (256K on K2 Thinking) | No | Modified MIT | Jul 2025 |
| DeepSeek V4 Pro | DeepSeek | 1.6T | 49B | MoE + sparse attention | 1M tokens | No | MIT | Apr 2026 |
| DeepSeek V4 Flash | DeepSeek | 284B | 13B | MoE + sparse attention | 1M tokens | No | MIT | Apr 2026 |
| DeepSeek V3.2 | DeepSeek | 671B | 37B | MoE | 128K tokens | No | MIT | Dec 2025 |
| DeepSeek R1 | DeepSeek | 671B | 37B | MoE | 128K tokens | No | MIT | Jan 2025 |
| GLM-5.2 | Z.ai (Zhipu) | 753B | ~40B | MoE | 1M tokens | No | MIT | Jun 2026 |
| GLM-5 | Z.ai (Zhipu) | 744B | 40B | MoE | 205K tokens | No | MIT | Feb 2026 |
| Qwen3.6-27B | Alibaba | 27B | 27B | Dense (hybrid attention) | 262K (1M via YaRN) | Text + Image + Video | Apache 2.0 | Apr 2026 |
| Qwen3.6-35B-A3B | Alibaba | 35B | 3B | MoE (hybrid attention) | 262K (1M via YaRN) | Text + Image + Video | Apache 2.0 | Apr 2026 |
| Qwen 3.5 397B-A17B | Alibaba | 397B | 17B | MoE | 256K tokens | Text + Image | Apache 2.0 | Feb 2026 |
| Qwen 3.5 122B-A10B | Alibaba | 122B | 10B | MoE | 256K tokens | Text + Image | Apache 2.0 | Feb 2026 |
| Qwen 3.5 27B | Alibaba | 27B | 27B | Dense | 256K tokens | Text + Image | Apache 2.0 | Feb 2026 |
| Qwen 3 235B | Alibaba | 235B | 22B | MoE (128e, 8 active) | 128K tokens | No | Apache 2.0 | Apr 2025 |
| Qwen 3 8B | Alibaba | 8B | 8B | Dense | 128K tokens | No | Apache 2.0 | Apr 2025 |
| gpt-oss-120b | OpenAI | 117B | 5.1B | MoE (native MXFP4) | 128K tokens | No | Apache 2.0 | Aug 2025 |
| gpt-oss-20b | OpenAI | 21B | 3.6B | MoE (native MXFP4) | 128K tokens | No | Apache 2.0 | Aug 2025 |
| Llama 4 Scout | Meta | 109B | 17B | MoE (16 experts) | 10M tokens | Text + Image | Llama 4 Community | Apr 2025 |
| Llama 4 Maverick | Meta | 400B | 17B | MoE (128 experts) | 1M tokens | Text + Image | Llama 4 Community | Apr 2025 |
| Llama 3.3 | Meta | 70B | 70B | Dense | 128K tokens | No | Llama 3.3 Community | Dec 2024 |
| Gemma 4 31B | 30.7B | 30.7B | Dense | 256K tokens | Text + Image | Apache 2.0 | Mar 2026 | |
| Gemma 4 26B A4B | 25.2B | 3.8B | MoE (128e, 8 active) | 256K tokens | Text + Image | Apache 2.0 | Mar 2026 | |
| Gemma 4 12B | 12B | 12B | Dense (encoder-free) | 256K tokens | Text + Image + Audio | Apache 2.0 | Jun 2026 | |
| Gemma 4 E4B | 8B | 4.5B effective | Dense (per-layer embeddings) | 128K tokens | Text + Image + Audio | Apache 2.0 | Mar 2026 | |
| Gemma 3 27B | 27B | 27B | Dense | 128K tokens | Text + Image | Gemma (agreement) | Mar 2025 | |
| Mistral Small 4 | Mistral AI | 119B | 6B | MoE (128e, 4 active) | 256K tokens | Text + Image | Apache 2.0 | Mar 2026 |
| Mistral Large 3 | Mistral AI | 675B | 41B | MoE | 256K tokens | Text + Image | Apache 2.0 | Dec 2025 |
| Phi-4 Reasoning Vision | Microsoft | 15B | 15B | Dense | 16K tokens | Text + Image | MIT | Mar 2026 |
| Phi-4 | Microsoft | 14B | 14B | Dense | 16K tokens | No | MIT | Jan 2025 |
| Phi-4 Mini | Microsoft | 3.8B | 3.8B | Dense | 128K tokens | No | MIT | Jan 2025 |
| Command A | Cohere | 111B | 111B | Dense | 256K tokens | No | CC-BY-NC | Mar 2025 |
| Falcon 3 10B | TII Abu Dhabi | 10B | 10B | Dense | 32K tokens | No | TII Falcon-LLM 2.0 | Dec 2024 |
Benchmark Scores
Benchmarks tell part of the story. MMLU-Pro tests professional-level knowledge, GPQA Diamond tests graduate-level reasoning, and SWE-bench Verified measures real-world coding ability against actual GitHub issues. All scores below come from official model cards or the developers’ published evaluations. Empty cells mean the score has not been published or independently verified. Watch the benchmark names carefully: SWE-bench Pro (used by Z.ai for GLM-5.2) is a harder, newer suite than SWE-bench Verified, so those numbers are not comparable across columns.
| Model | MMLU-Pro | GPQA Diamond | SWE-bench Verified | Other verified highlights |
|---|---|---|---|---|
| DeepSeek V4 Pro | N/A | N/A | 80.6% (max thinking) | Highest published open-weight SWE-bench Verified score |
| GLM-5 | N/A | N/A | 77.8% | Humanity’s Last Exam 50.4 |
| Qwen3.6-27B | N/A | N/A | 77.2% | Beats Qwen 3.5 397B MoE (76.2%) on the same suite |
| Qwen3.6-35B-A3B | 85.2% | 86.0% | 73.4% | 3B active params |
| gpt-oss-120b | N/A | 80.1% (high reasoning) | 62.4% (high reasoning) | 52.6% SWE-bench Verified at medium reasoning effort |
| GLM-5.2 | N/A | 91.2% | N/A | SWE-bench Pro 62.1, AIME 2026 99.2, Terminal-Bench 2.1 81.0 |
| DeepSeek R1 | 84.0% | 71.5% | N/A | MATH-500 97.3% |
| Qwen 3 235B | 83.8% | 77.1% | N/A | AIME ’24 85.7% |
A few things jump out. DeepSeek V4 Pro’s 80.6% on SWE-bench Verified is the strongest published coding result from any open-weight model, and the smaller V4 Flash inherits the same sparse-attention architecture at a fraction of the memory footprint. Qwen3.6-27B posting 77.2% is arguably the more remarkable number: that is a dense 27B model you can run on a single consumer GPU, beating Alibaba’s own 397B MoE flagship from two months earlier. On the aggregate Artificial Analysis Intelligence Index, Kimi K3 scores 57, the top open-weight result, with GLM-5.2 next at 51.
License Comparison
Licensing is where “open source” gets complicated. Some models are truly permissive (Apache 2.0, MIT), while others come with usage caps, geographic restrictions, or revenue gates. Read the fine print before building a product on any of these. The two biggest licensing stories this year: Google moved Gemma 4 to plain Apache 2.0, dropping the separate agreement Gemma 3 required, and Moonshot shipped Kimi K3 under a custom license that most people assumed was MIT and is not.
| License | Models | Commercial Use | Key Restrictions |
|---|---|---|---|
| Apache 2.0 | Qwen 3/3.5/3.6 (all open releases), gpt-oss-120b/20b, Gemma 4 (all sizes), Mistral Large 3, Mistral Small 4, Mixtral, Mistral 7B | Yes, unrestricted | None (gpt-oss adds a usage policy) |
| MIT | DeepSeek R1/V3.2/V4 Pro/V4 Flash, GLM-5, GLM-5.2, Phi-4 (all variants) | Yes, unrestricted | None |
| Modified MIT | Kimi K2 | Yes | Attribution required for very large commercial deployments |
| Kimi K3 License | Kimi K3 | Yes, with gates | Custom license, not MIT: model-as-a-service providers above $20M/year revenue need a separate agreement, and products above 100M MAU or $20M/month revenue must display “Kimi K3”. Internal use exempt |
| Llama 4 Community | Llama 4 Scout, Llama 4 Maverick | Yes, free under 700M MAU | EU multimodal restrictions; Meta license required above 700M monthly active users |
| Llama 3.3 Community | Llama 3.3 70B | Yes, free under 700M MAU | Same MAU threshold as Llama 4 |
| Gemma | Gemma 3 (all sizes) | Yes (requires agreement) | Must accept Google’s terms; Gemma 4 dropped this requirement entirely |
| CC-BY-NC | Command A, Command R+ | No | Non-commercial only; separate agreement required for commercial deployment |
| TII Falcon-LLM 2.0 | Falcon 3 (all sizes) | Yes | Apache 2.0-based with an acceptable use policy; no royalty or revenue gate |
If license flexibility is your top priority, the field has never been better: Qwen 3.6 and Gemma 4 under Apache 2.0, DeepSeek V4 and GLM-5.2 under MIT. You can fine-tune and deploy commercially with zero royalties. Even OpenAI’s gpt-oss models ship under Apache 2.0, though with a usage policy attached. The Kimi K3 license deserves a careful read if you are building a product on it, and the Llama licenses still carry the 700M MAU cap plus the EU multimodal exclusion.
Self-Hosting Resource Requirements
Benchmarks don’t tell you how a model feels when you’re actually running it. We tested six popular small models on a modest Ubuntu 24.04 VM (4 vCPUs, 16 GB RAM, CPU-only inference) using Ollama in March 2026. Each model answered the same prompt to keep things consistent.
| Model | Ollama Tag | Disk Size | RAM Usage | Response Time (CPU) | Notes |
|---|---|---|---|---|---|
| Llama 3.2 3B | llama3.2:3b | 2.0 GB | 11.4 GB | 88s | Clear, well-structured responses |
| Gemma 3 4B | gemma3:4b | 3.3 GB | 4.2 GB | 94s | Clean, structured, concise output |
| Phi-4 Mini 3.8B | phi4-mini | 2.5 GB | 8.9 GB | 97s | Good reasoning, occasional formatting artifacts |
| Mistral 7B | mistral:7b | 4.4 GB | 7.4 GB | 125s | Concise, accurate |
| Qwen 3 8B | qwen3:8b | 5.2 GB | 5.8 GB | 433s | Thinking mode adds latency; strong final answers |
| DeepSeek R1 8B | deepseek-r1:8b | 5.2 GB | 5.8 GB | 433s | Chain-of-thought reasoning; slow on CPU |
Gemma 3 4B stands out for RAM efficiency at just 4.2 GB, making it the best fit for memory-constrained environments. The reasoning models (Qwen 3 8B and DeepSeek R1 8B) took over 7 minutes each because their chain-of-thought process generates far more tokens before producing a final answer. On a GPU, those times drop to seconds. If you are choosing hardware for local inference, our GPU buyer guide for local LLMs breaks down which card fits each model size, and the VRAM requirements guide maps parameter counts to memory directly.
What the 2026 flagships actually need
The frontier open models are a different class of problem, so in August 2026 we measured two of them ourselves instead of quoting vendor claims.
| Model | Smallest usable quant | Memory floor | Measured speed | Hardware we tested on |
|---|---|---|---|---|
| DeepSeek V4 Flash | UD-IQ3_XXS, 103 GB | ~112 GB RAM (measured peak) | 5.61 tok/s generation, 12.33 tok/s prompt processing | 8-core CPU server, 128 GB RAM |
| Kimi K3 | UD-IQ1_S, 594 GB | 610 GB RAM + VRAM combined | 0.10 tok/s generation | 4x A100 40GB + 2 TB RAM |
| gpt-oss-120b | MXFP4 (ships natively quantized) | Single 80 GB GPU (vendor figure) | Not benchmarked by us | H100 or MI300X class |
Two findings from those runs are worth internalizing. First, DeepSeek V4 Flash is genuinely runnable on a 128 GB machine, and prompt processing, not generation, is the wall: at 12.33 tok/s prefill, a 10,000-token prompt takes over 13 minutes before the first output token appears. The full methodology and every measurement is in our DeepSeek V4 Flash local hardware test. Second, Kimi K3 is not practically self-hostable: one token every 10 seconds on four A100s, because 594 GB of weights turn the run into a memory-bandwidth problem no consumer budget solves. The numbers are in our Kimi K3 local run write-up. If a 100 GB class model is your target, the realistic home options are a 128 GB unified-memory box (see our Ryzen AI Max+ 395 mini PC comparison) or a multi-GPU workstation; for everything smaller, a mini PC built for local AI covers the 4B to 30B range comfortably.
Pull and run any of the small models with a single command:
ollama run gemma3:4bCheck our Ollama commands cheat sheet for the full list of management commands, and the Ollama models cheat sheet for current tags and quantization options across every family on this page.
Qwen 3, 3.5 and 3.6 (Alibaba)
Qwen is arguably the most versatile open-source model family available, and the release cadence is relentless: three generations in twelve months, all Apache 2.0. The Qwen 3 series (April 2025) spans from a 0.6B edge model to a 235B MoE flagship. Qwen 3.5 (February 2026) made the family natively multimodal, pushed context to 256K, and extended language coverage to 201 languages.
Qwen 3.6 (April 2026) is the generation that changed the economics. The two open releases pair a hybrid attention design (Gated DeltaNet linear attention mixed with standard self-attention) with a native 262K context extensible to roughly 1M tokens via YaRN scaling, and both accept text, images, and video. Qwen3.6-35B-A3B activates just 3B parameters per token yet posts 85.2% MMLU-Pro and 73.4% SWE-bench Verified. The dense Qwen3.6-27B is the headline act: 77.2% on SWE-bench Verified beats Alibaba’s own 397B MoE flagship, and the Q4_K_M quant is 16.8 GB, which fits a single consumer GPU. Both introduce Thinking Preservation, which carries reasoning traces across conversation turns instead of discarding them, a meaningful win for agent workloads.
The toggleable thinking mode remains the family’s signature feature. Enable chain-of-thought for math, logic, and coding; disable it for fast responses on straightforward queries. One model serves both use cases.
gpt-oss (OpenAI)
OpenAI shipping open weights was the event nobody had on their 2025 card. gpt-oss-120b and gpt-oss-20b, released August 2025 under Apache 2.0, are the company’s first open-weight models since GPT-2. Both are MoE reasoning models with a 128K context and three configurable reasoning effort levels (low, medium, high), and both ship natively quantized in MXFP4.
The efficiency story is the point. gpt-oss-120b activates only 5.1B of its 117B parameters per token and runs on a single 80 GB GPU, delivering near-parity with OpenAI’s o4-mini on core reasoning benchmarks and 62.4% on SWE-bench Verified at high reasoning effort. gpt-oss-20b activates 3.6B of 21B and runs on machines with 16 GB of memory, which puts a capable reasoning model on ordinary laptops. Both are text-only. A year on, community tooling support is excellent across Ollama, llama.cpp, and vLLM.
Kimi K2 and K3 (Moonshot AI)
Moonshot AI escalated the open-weights race twice. Kimi K2 (July 2025) is a 1T parameter MoE with 32B active and a Modified MIT license whose only catch is an attribution requirement for very large deployments; the original shipped with a 128K context, extended to 256K in later refreshes. It earned its reputation on agentic and tool-use workloads.
Kimi K3, whose weights went public on July 26, 2026, is the largest open model ever released: 2.8T total parameters, 104B active from 896 experts, native vision, video input, and a 1M token context. The weights arrive as 96 shards totaling about 1.56 TB on Hugging Face. Benchmark-wise it is the strongest open model on the aggregate indexes. Two caveats matter before you get excited. The license is a custom Kimi K3 document, not MIT: model-as-a-service providers above $20M/year revenue need a separate agreement, and very large products must display “Kimi K3” attribution, though internal use stays unrestricted. And self-hosting is effectively off the table: our own run needed 610 GB of combined RAM and VRAM for the smallest quant and generated at 0.10 tokens per second on four A100s. Full details in the Kimi K3 local run test. One operational gotcha from using the hosted version: K3 defaults to very high reasoning effort, and a low max-token budget can be consumed entirely by hidden reasoning, returning an empty response you still pay for. Budget 5,000+ output tokens or lower the reasoning effort.
DeepSeek V4, V3.2, and R1
DeepSeek V4 arrived in April 2026 as two MIT-licensed models sharing a sparse-attention architecture (token-wise compression plus DeepSeek Sparse Attention) and a 1M token default context: V4 Pro at 1.6T total parameters with 49B active, and V4 Flash at 284B total with 13B active. Both support thinking and non-thinking modes. A re-post-trained Flash build landed July 31, 2026: if you are downloading Flash weights, target the DeepSeek-V4-Flash-0731 repository on Hugging Face, since the original preview repo is superseded. V4 Pro still ships from its original repository.
V4 Pro’s 80.6% on SWE-bench Verified at max thinking is the best open-weight coding score published. V4 Flash is the one you can realistically run yourself: we measured 5.61 tokens per second generation on a CPU-only 128 GB box with the 103 GB UD-IQ3_XXS quant. Setup, expert-offload flags, and the prompt-processing bottleneck are covered in our hands-on DeepSeek V4 Flash guide.
The older generation still earns its keep. R1’s 97.3% on MATH-500 remains near the top, and the distilled versions (1.5B through 70B) put genuine chain-of-thought reasoning on consumer hardware; see running DeepSeek R1 locally with Ollama. V3.2 (December 2025) pioneered thinking integrated into tool-use workflows, and its Speciale variant achieved gold-medal performance at IMO 2025, IOI 2025, and the ICPC World Finals.
GLM-5 and GLM-5.2 (Z.ai)
Z.ai (formerly Zhipu AI) released GLM-5 in February 2026: 744B total parameters, 40B active, MIT license, trained entirely on 100,000 Huawei Ascend 910B chips with no US-manufactured hardware. Its 77.8% on SWE-bench Verified stood as the open-weight coding record for months.
GLM-5.2, launched June 13, 2026 with weights on Hugging Face days later, pushes to 753B total with roughly 40B active, a 1M token context, and the strongest open coding profile of any model on this page: 62.1 on SWE-bench Pro, 81.0 on Terminal-Bench 2.1, 99.2 on AIME 2026, and 91.2 on GPQA Diamond. Weights ship under MIT. Know what you are getting, though: GLM-5.2 is text-only. Z.ai’s vision line (GLM-5V-Turbo) is API-only with no downloadable weights, so if you need open-weight multimodal, look at Qwen 3.6, Gemma 4, or Kimi K3 instead.
Gemma 4 and Gemma 3 (Google)
Gemma 4 launched March 31, 2026 with four sizes, and fixed the two biggest complaints about the family in one release. The license is now plain Apache 2.0, no separate Google agreement required. And the lineup finally includes an MoE option: the 26B A4B activates 3.8B parameters from 128 experts, giving near-flagship quality at small-model inference cost. The encoder-free 12B followed in June 2026, rounding out the family to five sizes.
The sizes split cleanly by deployment target. E2B and E4B use per-layer embeddings to bring effective parameter counts (2.3B and 4.5B) below their raw sizes, run a 128K context, and take text, image, and audio input; they are the phone and edge-device models. The 12B has the interesting architecture: encoder-free, projecting image and audio directly into the transformer’s embedding space instead of bolting on separate encoders. The 26B A4B and 31B dense flagships run a 256K context with text and image input. All sizes support function calling, structured JSON output, and native system instructions, and training covered over 140 languages.
Gemma 3 remains relevant for one reason: it is everywhere, and it is proven. Our March CPU tests above show why the 4B stays popular in tight-memory deployments. New projects should start on Gemma 4.
Llama 4 (Meta)
Meta’s Llama 4 family (April 2025) brought MoE to the Llama line. Scout (109B total, 17B active) carries a 10 million token context window, still the longest of any open model by a wide margin. Maverick (400B total, 17B active, 128 experts) targets production deployments with a 1M context. Both are natively multimodal.
The family has been strangely static since. Behemoth, the roughly 2T parameter flagship, remains unreleased as of August 2026 after Meta paused it in 2025 over incremental internal results. Periodic reports of a “Llama 4.5” or “Llama 5” keep circulating, and none have materialized on any first-party Meta channel; treat specific spec claims you read about them as fabricated until Meta says otherwise. Meanwhile the licensing catch is unchanged: the Llama 4 Community License is free under 700M MAU, but the Acceptable Use Policy excludes multimodal rights for EU-based individuals and companies, which effectively restricts the whole Llama 4 family in the EU. Llama 3.3 70B (dense, text-only) is unaffected and remains popular because it fits a single high-end GPU without MoE-aware serving infrastructure.
Mistral Models
Mistral’s current pair covers both ends of the spectrum under Apache 2.0. Mistral Large 3 (December 2025) is a 675B MoE with 41B active, multimodal across 80+ languages. Mistral Small 4 (March 2026) is the clever one: 119B total with only 6B active through 4-of-128 expert routing, unifying instruction following, configurable-depth reasoning, and multimodal input in one model. It absorbed what used to be three separate product lines (Magistral for reasoning, Pixtral for vision, Devstral for agentic coding).
A new open-weight family is coming: Mistral confirmed in early July 2026 that a larger “fat but sparse” MoE family entered early access with research and government partners, with no public parameter count, benchmarks, or ship date yet. The original Mistral 7B and Mixtral 8x7B remain in wide deployment and are still serviceable Apache 2.0 fallbacks on old hardware, though every current alternative outperforms them.
Phi-4 (Microsoft)
Microsoft’s Phi-4 family proves that smaller models can punch above their weight on specific tasks. The 14B base model and its reasoning variants excel at math and logic, consistently outperforming larger models there. Phi-4 Mini at 3.8B with a 128K context remains one of the best options for resource-constrained deployments that need long context, though in our testing it used 8.9 GB of RAM, more than expected for its size, likely due to context window allocation.
Phi-4 Reasoning Vision (15B, March 2026) adds image understanding to the reasoning pipeline via a SigLIP-2 vision encoder, with dynamic resolution up to 3,600 visual tokens for GUI grounding and document analysis. It decides per query whether deep reasoning is warranted, saving compute on simple questions. All Phi-4 variants ship under MIT.
Command A, Falcon 3, and the Rest
Command A (Cohere, 111B dense) offers a 256K context, runs on two A100/H100 GPUs, and stays optimized for retrieval-augmented generation and tool use. The CC-BY-NC license limits it to non-commercial use without a separate agreement, which keeps it out of most production shortlists despite the solid engineering. Cohere’s smaller releases, Tiny Aya (3.35B, 70+ languages, CC-BY-NC) and the Apache 2.0 Cohere Transcribe speech model, fill niches the big families ignore.
Falcon 3 from TII Abu Dhabi (1B to 10B, trained on 14T tokens) ships under the TII Falcon-LLM License 2.0, an Apache 2.0-based license with an acceptable use policy and free commercial use. DBRX and Grok-1, both early-2024 MoE releases, have aged out of practical relevance, and xAI’s promised Grok 3 open-sourcing has still not produced published weights as of August 2026.
Which Model Should You Choose?
The “best” model depends entirely on what you’re building. Here’s a quick decision guide based on the verified numbers above and our own runs.
Best for coding: DeepSeek V4 Pro (80.6% SWE-bench Verified) if you have the serving infrastructure, GLM-5.2 for terminal and agentic work (81.0 Terminal-Bench 2.1). On a single consumer GPU, Qwen3.6-27B at 77.2% SWE-bench Verified is the clear pick.
Best for reasoning and math: GLM-5.2 (99.2 AIME 2026, 91.2 GPQA Diamond) or Kimi K3 (top aggregate open-model scores). DeepSeek R1 remains excellent for pure math on smaller budgets.
Best on a single 80 GB GPU: gpt-oss-120b was built for exactly this slot and owns it.
Best for edge and laptops: gpt-oss-20b if you want reasoning in 16 GB; Gemma 4 E2B/E4B if you want image and audio input on-device; Phi-4 Mini for long context in under 4 GB of weights.
Best for multimodal: Qwen 3.6 (text, image, video, open weights) or Gemma 4 (adds audio on the 12B and E-series). GLM-5.2 and gpt-oss are text-only; remember that before committing.
Best for long context: Llama 4 Scout’s 10M tokens is still unmatched on paper. For a practical 1M with current-generation quality, Kimi K3, DeepSeek V4, GLM-5.2, or Qwen 3.6 via YaRN.
Best for multilingual: Qwen 3.5/3.6 (201 languages) or Gemma 4 (140+ languages, Apache 2.0).
Best permissive license: Qwen 3.6, Gemma 4, gpt-oss (Apache 2.0), DeepSeek V4, GLM-5.2, Phi-4 (MIT). No usage caps, no royalties, no geographic restrictions. Kimi K3 and the Llama family both carry gates worth reading twice.
Running These Models Locally with Ollama
Every model in the March test table, plus gpt-oss and the Qwen 3.6 releases, can be pulled and run with a single Ollama command. Install Ollama first:
curl -fsSL https://ollama.com/install.sh | shThen pull and interact with any model:
ollama run gpt-oss:20bOn a GPU-equipped machine, Ollama automatically uses CUDA or ROCm, cutting response times from minutes to seconds. On CPU-only systems, stick with models under 8B parameters for usable response times, or step up to a unified-memory box for the 100 GB class; our Mac mini vs mini PC vs GPU comparison walks through that decision.
For a full setup walkthrough, see our guide on installing Ollama on Rocky Linux and Ubuntu. If you want a ChatGPT-style web interface for your local models, Open WebUI provides exactly that. And when Ollama’s abstractions get in the way of squeezing out performance, running llama.cpp directly buys you measurably higher throughput on the same hardware.
