LLM price barometer

Changes of 1% or more in the price we quote, from 14 Nov to 21 Nov, biggest first. Percentages are on the blended price (30% input, 70% output). Data snapshot: 2026-11-21.

Past weeks · RSS feed

7 price changes (4 up, 3 down) among 306 models compared. 24 other moves are flagged as a change of quoted host or an outlier listing.

Price changes (7)

Price increases (4)

Price decreases (3)

Quoted host changes and outliers (24)

Many models are sold by several hosts at different rates, and we quote one of them. These moves are flagged as a change of quoted host: the new rate matches a different host from the one we quoted at the start of the window. Rows marked Outlier listing involve a quoted rate far from the model's typical host rate.

Price increases (7)

Price decreases (17)

Since tracking began

Average quoted price

Average quoted price since 2026-08-15, first day = 100: now 101.7.9010011015 Aug9 Sep3 Oct28 Oct21 Nov101.7
Average change in the price we quote for the same 189 models, 15 Aug = 100. 148 are quoted at the same price as on 15 Aug, 25 higher and 16 lower.

Median price per 1M tokens

Median quoted price per 1M tokens since 2026-08-15: $1.43 then, $1.39 now; middle half $0.38 to $4.80.$0$1$2$3$4$515 Aug9 Sep3 Oct28 Oct21 Nov$1.39
Median price we quote per 1M tokens (30% input, 70% output) across every model listed that day: $1.43 for 240 models on 15 Aug, $1.39 for 308 on 21 Nov. The band is the middle half of them, now $0.38 to $4.80 ($0.46 to $3.41 on 15 Aug). Unlike the average above, it moves when models join or leave the catalogue, even if no price changes.

Models added and removed

Models added and removed by week: 254 priced on 2026-08-15, 316 on 2026-11-21.AddedRemoved+11-431 Aug+7-27 Sep+314 Sep+20-321 Sep+3-128 Sep+4-105 Oct+35-412 Oct+4-219 Oct+4-1326 Oct+8-92 Nov+9-69 Nov+10-616 Nov*
Priced models added to and removed from the catalogue, by week starting Monday. 254 on 15 Aug, 316 on 21 Nov: 132 added, 70 removed. * Week in progress.

Every price change, day by day

The price we quote changed at least once for 88 of 386 models between 30 Aug and 21 Nov: 676 changes, most-changed model first. 390 are established as a change of quoted host (outlined).

Every change to a quoted LLM price since 2026-08-30: 88 models, 676 changes.UpDownQuoted host changedListed31 Aug7 Sep14 Sep21 Sep28 Sep5 Oct12 Oct19 Oct26 Oct2 Nov9 Nov16 NovDeepSeek DeepSeek V4.1 Flash: 38 changesDeepSeek V4.1 FlashDeepSeek DeepSeek V4 Flash 0423: 36 changesDeepSeek V4 Flash 0423DeepSeek DeepSeek V4 Pro 0423: 36 changesDeepSeek V4 Pro 0423DeepSeek DeepSeek V4 Flash 0731: 35 changesDeepSeek V4 Flash 0731DeepSeek DeepSeek V4 Pro 0813: 30 changesDeepSeek V4 Pro 0813Google Gemma 4 26B A4B: 30 changesGemma 4 26B A4BZhipu GLM 5.3 Flash: 28 changesGLM 5.3 FlashZhipu GLM 5.2: 27 changesGLM 5.2Moonshot Kimi K3: 26 changesKimi K3Zhipu GLM 5.3: 22 changesGLM 5.3Alibaba Qwen3.8 27B: 17 changesQwen3.8 27BZhipu GLM 5.1: 15 changesGLM 5.1Meta Llama 3.3 70B Instruct: 13 changesLlama 3.3 70B InstructAlibaba Qwen3.6 27B: 12 changesQwen3.6 27BAlibaba Qwen3.5-35B-A3B: 11 changesQwen3.5-35B-A3BMoonshot Kimi K2.7 Code: 11 changesKimi K2.7 CodeNvidia Nemotron 3 Ultra: 11 changesNemotron 3 UltraAlibaba Qwen3.5 397B A17B: 10 changesQwen3.5 397B A17BAlibaba Qwen3.5-9B: 10 changesQwen3.5-9BNvidia Nemotron 3.5 Lightning: 10 changesNemotron 3.5 LightningTencent Hy3: 10 changesHy3Alibaba Qwen3 30B A3B Instruct 2507: 9 changesQwen3 30B A3B Instruct 2507OpenAI gpt-oss-120b: 9 changesgpt-oss-120bDeepSeek DeepSeek V3: 8 changesDeepSeek V3ibm-granite Granite 4.2 8B: 8 changesGranite 4.2 8Bmeta Muse Glimmer 30B: 8 changesMuse Glimmer 30BMiniMax MiniMax M2.5: 8 changesMiniMax M2.5Alibaba Qwen3.5-27B: 7 changesQwen3.5-27BDeepSeek DeepSeek V4 Flash Vision Exp: 7 changesDeepSeek V4 Flash Vision ExpOpenAI gpt-oss-20b: 7 changesgpt-oss-20bZhipu GLM 4.6: 7 changesGLM 4.6DeepSeek DeepSeek V3 0324: 6 changesDeepSeek V3 0324DeepSeek DeepSeek V4.1 Flash 1017: 6 changesDeepSeek V4.1 Flash 1017Google Gemma 3 27B: 6 changesGemma 3 27BMeta Llama 3.2 3B Instruct: 6 changesLlama 3.2 3B InstructMiniMax MiniMax M2.7: 6 changesMiniMax M2.7Alibaba Qwen3 235B A22B Instruct 2507: 5 changesQwen3 235B A22B Instruct 2507Alibaba Qwen3.6 35B A3B: 5 changesQwen3.6 35B A3BDeepSeek DeepSeek V3.2: 5 changesDeepSeek V3.2Meta Llama 4 Maverick: 5 changesLlama 4 MaverickMoonshot Kimi K2.6: 5 changesKimi K2.6Nvidia Nemotron 3 Super: 5 changesNemotron 3 SuperAlibaba Qwen3 14B: 4 changesQwen3 14BMiniMax MiniMax M1: 4 changesMiniMax M1Moonshot Kimi K2.5: 4 changesKimi K2.5Moonshot Kimi K3.1 Pro: 4 changesKimi K3.1 ProNvidia Nemotron 3.5 Super VL: 4 changesNemotron 3.5 Super VLreflection Beam 1.0 501B A23B: 4 changesBeam 1.0 501B A23BXiaomi MiMo-V2.5-Pro: 4 changesMiMo-V2.5-ProZhipu GLM 4.7: 4 changesGLM 4.7Alibaba Qwen3 Next 80B A3B Instruct: 3 changesQwen3 Next 80B A3B InstructAlibaba Qwen3.5-122B-A10B: 3 changesQwen3.5-122B-A10BDeepSeek DeepSeek V3.1: 3 changesDeepSeek V3.1Meta Llama 3.1 70B Instruct: 3 changesLlama 3.1 70B InstructMeta Llama 3.1 8B Instruct: 3 changesLlama 3.1 8B InstructMeta Llama 4 Scout: 3 changesLlama 4 ScoutMistral Mistral Small 3.2 24B: 3 changesMistral Small 3.2 24BNvidia Nemotron 3 Nano 30B A3B: 3 changesNemotron 3 Nano 30B A3Bthinkingmachines Inkling: 3 changesInklingAlibaba Qwen3 30B A3B: 2 changesQwen3 30B A3BAlibaba Qwen3 Coder 30B A3B Instruct: 2 changesQwen3 Coder 30B A3B InstructDeepSeek DeepSeek V3.1 Terminus: 2 changesDeepSeek V3.1 TerminusGoogle Gemma 3 4B: 2 changesGemma 3 4Binclusionai Ling 3.0 Flash VL: 2 changesLing 3.0 Flash VLinclusionai Ling 3.1 Flash: 2 changesLing 3.1 FlashMiniMax MiniMax M2: 2 changesMiniMax M2Mistral Mistral Large 4: 2 changesMistral Large 4particle-ai Particle-1: 2 changesParticle-1upstage Solar Pro 4: 2 changesSolar Pro 4Xiaomi MiMo-V2.5: 2 changesMiMo-V2.5Zhipu GLM 4.5 Air: 2 changesGLM 4.5 AirAlibaba Qwen3 235B A22B Thinking 2507: 1 changeQwen3 235B A22B Thinking 2507Alibaba Qwen3 VL 30B A3B Thinking: 1 changeQwen3 VL 30B A3B ThinkingAlibaba Qwen4 Max: 1 changeQwen4 Maxarcee-ai Trinity Large Thinking: 1 changeTrinity Large ThinkingDeepSeek R1 0528: 1 changeR1 0528Google Gemma 4 31B: 1 changeGemma 4 31Binclusionai Ling 3.0 Flash Fin: 1 changeLing 3.0 Flash FinMistral Devstral 2 2512: 1 changeDevstral 2 2512Mistral Mistral Medium 3: 1 changeMistral Medium 3Mistral Mistral Medium 3.1: 1 changeMistral Medium 3.1Mistral Mistral Nemo: 1 changeMistral Nemoricursiveai Ripple 65B Preview: 1 changeRipple 65B PreviewTencent Hy4 preview: 1 changeHy4 previewxAI Grok 4.7: 1 changeGrok 4.7Zhipu GLM 4.7 Flash: 1 changeGLM 4.7 FlashZhipu GLM 5: 1 changeGLM 5Zhipu GLM 5.3 Prime: 1 changeGLM 5.3 Prime

How these figures are calculated

The full cost-per-request formulas are written out in the cost per request methodology guide.

What to do with a price move

Cloud FinOps skill on GitHub

LLM pricing · Models by maker · Cloud compute pricing · Compare models · Price barometer · FinOps guides · API and MCP server