LLM costs: start here.
Most LLM bills are decided by three things: how much context you re-send, whether it hits the cache, and which model tier you chose. Start with how pricing works, then run your own numbers in the calculator.
Kept current
LLM API cost calculator
Enter your requests and token counts and compare the monthly bill across every model in the pricing tracker.
Open →Tracker · updated Nov 12, 2026LLM API Pricing Tracker
A living tracker of per-million-token API list prices for OpenAI, Anthropic, Google Gemini, Mistral, xAI and DeepSeek: input, cached input, output, context window and scheduled price changes, each row linked to the vendor page and kept current.
Open →Tracker · updated Nov 12, 2026AI Model Release Tracker
A living tracker of notable large language model releases since July 2026, hosted and open-weight: release date, vendor, context window and licence or access terms, each row linked to a primary source and kept current.
Open →Read in this order
- LLM Tokens and Pricing Explained (2026): $0.10 to $50 per Million
What an LLM token is, why tokenizers count the same text differently, how input, output, cached and reasoning tokens are billed, the November 2026 per-million prices for OpenAI, Claude and Gemini, and a script to estimate your monthly bill.
- Prompt Caching Explained (2026): 0.05x Reads, 5-Minute TTLs, Real Math
How prompt caching works in 2026: KV-cache prefix matching, Anthropic vs OpenAI vs Gemini rules, TTLs from 5 minutes to 24 hours, write premiums of 1.25x to 2.5x, reads at 0.025x to 0.1x, batch stacking, and a worked bill that falls from $6,700 to $1,002.
- FinOps for AI in 2026: 98% Manage Token Spend, 3 in 4 Can't Prove the Value, GPUs Run at 5%, and FOCUS Gets a Token Column in 1.5
The AI cost ledger as of October 2026: State of FinOps 2026 (1,192 practitioners, 98 percent manage AI spend, up from 31 percent in 2024), the Tokenomics Foundation finding that three in four enterprises cannot prove AI outcomes to the CFO, a per-million-token price table for Anthropic, OpenAI, Google and DeepSeek with cache and batch multipliers, a 20-step agent loop at $1.58 uncached and $0.42 cached, the FOCUS 1.2 to 1.5 roadmap and who exports which version, LiteLLM and Cloudflare budgets, Cast AI's 5 percent GPU utilisation, and $0.0037 of model cost inside a $0.99 resolution.
- RAG vs Long Context vs Fine-Tuning in 2026: 1M Tokens Isn't Free
RAG vs long context vs fine-tuning in 2026: current 1M-token context windows and prices, what context rot, NoLiMa and RULER show about long-context accuracy, where RAG fails, the per-query cost math, and a decision framework you can apply this quarter.
- Self-Hosting LLMs vs API Costs in 2026: The 62% Utilisation Break-Even
Self-hosting an open LLM vs paying per token in October 2026: H100 prices from $1.17 to $4.29 an hour, 8,973 tokens/s per GPU for gpt-oss-120b on vLLM, a $0.20 per million API price, the utilisation each price needs to break even, and the ops costs the spreadsheet leaves out.
- AI Gateways in 2026: Six Options Compared After LiteLLM's 40-Minute Backdoor
What an AI gateway (LLM gateway) does in 2026 and how LiteLLM, Portkey, Kong, Cloudflare, Envoy AI Gateway (now Agent Router) and OpenRouter compare on routing, budgets, caching and guardrails, plus the LiteLLM PyPI backdoor, three LiteLLM CVEs on CISA KEV, and build vs buy.
- LLM VRAM Requirements in 2026: Weights, KV Cache and Which GPU Fits
How much VRAM an LLM needs in 2026: weight memory at BF16, FP8 and 4-bit, the KV cache formula with real config.json numbers, why context and batch size dominate, and memory specs for RTX 4090/5090, L40S, H100, H200, B200, MI300X and MI355X.
- GPU Clouds in 2026: H100s at $2.59 an Hour, $104B of CoreWeave Backlog on $35B of Debt, and a Six-Year Bet on Silicon
The GPU rental economy from filings and price pages fetched on 5 October 2026: H100 hours that fell from about $8 in 2023 to a $1.70 trough in October 2025 and recovered to $2.59 on the transacted index while list prices rose 25 percent; CoreWeave's $2.58B quarter, $103.7B backlog, $35.1B of debt from 15 percent DDTLs to a 5.9 percent A3 facility, and Microsoft at 67 percent of 2025 revenue; Nebius, Lambda, Crusoe, Oracle's $664B RPO and NVIDIA's $105B of guarantees; who books servers over 5, 5.5 or 6 years and what Burry's $176B means; and an own-versus-rent break-even calculator.
Guides and fixes
- How to Cut LLM API Costs: Prompt Caching, Batching, Routing (2026)
- LiteLLM Proxy Setup: Docker, Virtual Keys, Budgets (2026)
- How to Fix LLM API 429 Rate Limit and 529 Overloaded Errors (2026)
- How to Calculate VRAM and KV Cache for LLM Inference (2026)
- How to Deploy vLLM on Kubernetes with GPUs: Step-by-Step (2026)
- How to Benchmark LLM Inference: TTFT, Throughput and Goodput (2026)
- How to Run a Local LLM With Ollama: Setup, GPU and API Guide (2026)
Latest on LLM costs
How to Quantize an LLM to GGUF with llama.cpp: Q4_K_M and imatrix (2026)
Step-by-step llama.cpp b12110 guide: download a Hugging Face model, convert it with convert_hf_to_gguf.py, build an importance matrix, quantize to Q4_K_M, Q5_K_M and Q8_0, measure perplexity and KL divergence, and serve it with llama-server.
11 minNov 10, 2026LLM Quantization Explained (2026): FP8, NVFP4, GGUF, AWQ and GPTQ
What LLM quantization does to weights, activations and the KV cache, how FP8, INT4, MXFP4, NVFP4 and GGUF k-quants differ, what the published quality loss looks like, and which format to use with llama.cpp, vLLM, TensorRT-LLM and MLX.
12 minNov 10, 2026Gemini 4 Argon and Carbon API Prices Double to $4/$20 on 1 December
Google made Gemini 4 Argon available to developers on 13 October 2026 at $2/$10 per million tokens, a promo that ends on 30 November. Gemini 4 Carbon followed for cyber defenders on 28 October and went generally available on 10 November at the same prices. The prices, benchmarks and what to budget.
7 minNov 10, 2026Claude Fable 5.2 Launches at $8/$40, 20% Below Fable 5.1, With Mythos 5.2
On 9 November 2026 Anthropic released Claude Fable 5.2 and Mythos 5.2, one model with two safeguard profiles. Prices fell 20% to $8/$40 per million tokens. The benchmarks, where rivals still win, and what it means for engineering teams.
7 minOct 31, 2026How to Install NVIDIA GPU Operator and Share GPUs: Time-Slicing, MIG (2026)
Step-by-step NVIDIA GPU Operator v26.7.2 install with Helm, verified with nvidia-smi and a CUDA test pod, then GPU sharing with a time-slicing ConfigMap or MIG profiles, DCGM Exporter metrics, and fixes for driver crash loops and nodes with no nvidia.com/gpu.
14 minOct 31, 2026LLM Inference Engines 2026: vLLM vs SGLang vs TensorRT-LLM vs Dynamo
vLLM 0.32, SGLang 0.5.22, TensorRT-LLM 1.2.1 stable (1.4 in RC), Dynamo 1.6 and llama.cpp compared as of October 2026, with TGI now archived. How batching, PagedAttention and speculative decoding differ, what published benchmarks show, and which engine to pick.
11 minOct 31, 2026GPUs on Kubernetes Explained (2026): Device Plugin, MIG, Time-Slicing, DRA
How GPUs on Kubernetes work in October 2026: the device plugin and nvidia.com/gpu, what GPU Operator v26.7.2 installs, MIG vs time-slicing vs MPS, DRA (GA since 1.34) and NVIDIA's DRA driver v0.5.1, Kueue quotas, EKS, GKE and AKS options, and why GPUs average 5% use.
14 minOct 20, 2026Running LLMs Locally in 2026: Ollama vs llama.cpp vs LM Studio vs MLX
A practical 2026 guide to running LLMs locally: Ollama 0.40, llama.cpp 0.7, LM Studio 0.4, MLX and vLLM compared, which models fit in 8, 16, 32 and 64GB, how Q4_K_M quantization works, the bandwidth maths behind tokens per second, licences and privacy.
12 min