{"slug": "glm-5-2-cheapest-inceptron-0-90-m-input", "title": "GLM 5.2 - cheapest: Inceptron $0.90/M input", "summary": "GLM 5.2, a 744B-parameter Mixture-of-Experts model with ~40B active parameters per token, was released in June 2026 under an MIT license. It uses MLA and DeepSeek Sparse Attention for a 1M context window and is described as the strongest open-source coding/reasoning model as of its release date. Cloud pricing starts at $0.90 per million input tokens via Inceptron.", "body_md": "# GLM 5.2\n\nMoE workstation744B total, ~40B active per token (MoE: 256 routed experts, 8 active + 1 shared). Uses MLA + DeepSeek Sparse Attention (IndexShare) for a solid 1M context. Q4_K_M (~410GB) fits a 512GB Mac Studio M4 Ultra or 4x DGX Spark; Q2_K (~240GB) fits 2-3 DGX Sparks. Strongest open-source coding/reasoning model as of June 2026.\n\n- 744.0B\n- 1000k\n- mit\n- Jun 2026\n\n1\nperson run this as their daily driver.\n[See the leaderboard](/daily-drivers)\n\n## Scores\n\n## Run it locally\n\nPer-quant memory needs and a static \"can you run it?\" reference - no rig entry required\n\n### Can you run it? - reference rigs\n\n| Rig | Q2_K | Q3_K_M | Q4_K_M | Q5_K_M | BF16 |\n|---|---|---|---|---|---|\n| NVIDIA Jetson Orin NX 16GB |\n|\n\n[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)[no -> cloud](#cloud-pricing)Fit tiers use the same will-it-run logic as the rig finder. For comfortable fits, the badge reflects decode speed: fast >=20 t/s, ok 8-20 t/s, slow <8 t/s. t/s is a bandwidth estimate, not a measured benchmark.\n\n**How can a 24GB GPU run a 744B model?** It does not load the model into VRAM. The quantized weights (e.g. ~410GB at Q4) sit in system RAM; the GPU only holds the small shared attention and router tensors and accelerates prompt processing. Because GLM 5.2 is a Mixture-of-Experts model, each token activates only ~40B of its 744B params, so llama.cpp streams just those active experts from system RAM to the GPU each token (the `-cmoe`\n\noffload path).\n\nThat makes **decode speed bound by system-RAM bandwidth, not GPU bandwidth** - single digits on DDR4, which is why these rigs show 3-8 t/s even though they “fit.” A bigger GPU (e.g. 2x 3090) keeps more experts resident on-card and raises tok/s; a smaller GPU still runs it but pays the bandwidth tax. A 744B dense model could not run this way - only MoE’s small-active-params trick makes it possible.\n\nAggressive quants (1-2 bit) trade accuracy for size - roughly 17% accuracy loss at 2-bit vs full precision, and real long-context work often needs Q5 or Q6 even when lower quants “fit.”\n\nFormula estimates here are conservative; real tuned setups can exceed them (one HN user reports ~6 tok/s on a 512GB DDR4 + 2x 3090 rig).\n\n## Download options\n\n## Or run it in the cloud\n\nLive per-provider pricing, throughput and uptime - refreshed about 10 hours ago via OpenRouter. Click a column to sort.\n\n| Provider | Type | Input $/M | Output $/M | Cache $/M | Tok/s | Latency | Uptime | Value |\n|---|---|---|---|---|---|---|---|---|\n|\nWandB\n|\nAPI | 1.39 | 4.40 | 0.260 | - | - | 100.00% | best uptime |\n| API | 1.40 | 4.40 | - | - | - | - | ||\n| API | 3.00 | 10.25 | 0.500 | - | - | 100.00% | ||\n|\nAlibaba\n|\nAPI | 1.32 | 4.14 | 0.264 | - | - | 99.97% | |\n|\nNovita\n|\nAPI | 0.98 | 3.08 | 0.182 | - | - | 99.95% | |\n|\nFireworks\n|\nAPI | 2.10 | 6.60 | 0.210 | - | - | 99.94% | |\n|\nTogether\n|\nAPI | 1.40 | 4.40 | 0.260 | - | - | 99.88% | |\n|\nStreamLake\n|\nAPI | 1.12 | 3.52 | 0.208 | - | - | 99.87% | |\n|\nAtlasCloud\n|\nAPI | 1.26 | 3.96 | 0.234 | - | - | 99.84% | |\n| API | 1.40 | 4.40 | 0.260 | - | - | 99.78% | ||\n|\nBaidu\n|\nAPI | 0.97 | 3.07 | 0.181 | - | - | 99.73% | |\n|\nAkashML\n|\nAPI | 1.30 | 4.40 | 0.180 | - | - | 99.72% | |\n|\nDecart\n|\nAPI | 1.20 | 4.20 | 0.200 | - | - | 99.71% | |\n|\nSiliconFlow\n|\nAPI | 1.30 | 4.09 | 0.260 | - | - | 99.64% | |\n|\nAmbient\n|\nAPI | 1.40 | 4.40 | 0.260 | - | - | 99.51% | |\n|\nZ.AI\n|\nAPI | 1.40 | 4.40 | 0.260 | - | - | 99.47% | |\n|\nIo Net\n|\nAPI | 1.60 | 4.99 | 0.799 | - | - | 99.25% | |\n|\nDeepInfra\n|\nAPI | 0.93 | 3.00 | 0.180 | - | - | 98.84% | |\n|\nParasail\n|\nAPI | 1.40 | 4.40 | 0.260 | - | - | 98.81% | |\n|\nVenice\n|\nAPI | 1.40 | 4.40 | 0.260 | - | - | 98.73% | |\n|\nCloudflare\n|\nAPI | 1.40 | 4.40 | 0.260 | - | - | 98.47% | |\n|\nInceptron\n|\nAPI | 0.90 | 3.08 | 0.180 | - | - | 98.04% | cheapest |\n|\nDigitalOcean\n|\nAPI | 1.05 | 4.40 | 0.210 | - | - | 97.56% | |\n| API | 0.98 | 3.08 | 0.182 | - | - | 97.15% | ||\n| API | 1.40 | 4.40 | 0.700 | - | - | 95.48% | ||\n|\nMorph\n|\nAPI | 1.10 | 4.10 | 0.180 | - | - | 95.32% | |\n|\nDekaLLM\nrisky\n|\nAPI | 0.94 | 3.00 | 0.180 | - | - | 92.01% | |\n| Sub | - | - | - | - | - | - | $10.00/mo Coding Plan Lite | |\n| Sub | - | - | - | - | - | - | $20.00/mo Pro | |\n| Sub | - | - | - | - | - | - | $30.00/mo Coding Plan Pro | |\n| Sub | - | - | - | - | - | - | $80.00/mo Coding Plan Max | |\n| Sub | - | - | - | - | - | - | $100.00/mo Max |\n\nDefault order: throughput among 95%+ uptime providers, then latency; subscriptions last. Sort by any column. Subscription rows show $/mo in the Value column - per-token columns are \"-\". Affiliate links are marked sponsored / nofollow. Confirm current pricing on the provider's site before committing.\n\n[Detailed API pricing page + JSON endpoint →](/models/glm-5-2/pricing)\n\n## Inference cost over time\n\nData accumulates from the first daily sync - longer ranges populate over time. Prices come from OpenRouter snapshots, not a historical API.", "url": "https://wpnews.pro/news/glm-5-2-cheapest-inceptron-0-90-m-input", "canonical_source": "https://tokenstead.ai/models/glm-5-2", "published_at": "2026-07-09 14:12:02+00:00", "updated_at": "2026-07-09 15:10:15.095978+00:00", "lang": "en", "topics": ["large-language-models", "artificial-intelligence", "ai-products"], "entities": ["GLM 5.2", "Inceptron", "DeepSeek", "WandB", "OpenRouter"], "alternates": {"html": "https://wpnews.pro/news/glm-5-2-cheapest-inceptron-0-90-m-input", "markdown": "https://wpnews.pro/news/glm-5-2-cheapest-inceptron-0-90-m-input.md", "text": "https://wpnews.pro/news/glm-5-2-cheapest-inceptron-0-90-m-input.txt", "jsonld": "https://wpnews.pro/news/glm-5-2-cheapest-inceptron-0-90-m-input.jsonld"}}