# AI cost per token in 2026: real cost, caching and routing Source: https://slash-digital.io/en/insights/ai-costs/ Strategy and regulation · Updated October 2026 Price per million tokens is only the start. The real bill depends on caching, reasoning effort, retries and architecture decisions made before the first prompt is written. Here is how we work it out. - $10 / $50 new premium tier per million tokens - 7.7x cost per task, max vs low effort on Opus 5.5 - 50% batch discount at Anthropic, OpenAI and Google Start a project All insights Esteban Zuleta, Co-founder, business and finance, Slash Digital Lab Published: September 26, 2026 Updated: October 28, 2026 Reading time: 9 min Topic: Strategy and regulation In short In 2026 AI prices moved two ways: a premium tier at $10 / $50 per million tokens appeared (Claude Fable 5.1, GPT-6 Astra) while mid-tier prices fell (GPT-6.1 Sol at $2 / $10, Claude Opus 5.5 20% below Opus 5). The bill, though, depends less on list prices than on caching, reasoning tokens, batch discounts and surcharges. Our rule: measure cost per solved task, not price per token, apply the levers in order and run the local-versus-API numbers with your own volumes before buying hardware. For your company Choosing a model or an AI architecture for your company? Slash Digital Lab helps you decide with your own data, costs and constraints. AI consulting → ## How AI is billed in 2026 Eight meters run on every invoice; most teams watch two. | Meter | How it works | Examples | | --- | --- | --- | | Input and output | Output costs five to six times input | Claude Opus 5.5 $4 / $20; GPT-6.1 Sol $2 / $10; Gemini 3.8 Flash $0.75 / $3.75 | | Reasoning tokens | Billed as output, even when you only see a summary | OpenAI and Google document it; effort sets the volume | | Cached input | Reused prefixes at a fraction of input; writing the cache costs extra | Reads at 10% of input at Anthropic, on OpenAI's GPT-5 family, GPT-6 Sol, GPT-6 Astra and GPT-6 Luna and on Gemini 3.8 Flash; 5% on Claude Opus 5.5 and Sonnet 5.5 (since October 7), GPT-6.1 Sol, GPT-6.1 Terra and Gemini 4 Argon; 2.5% on Claude Fable 5.1; Anthropic writes at 1.25x or 2x | | Batch | Asynchronous jobs at half price | 50% off at Anthropic, OpenAI and Google | | Speed | Premium for faster responses | OpenAI Fast (called Priority until July 30, 2026) at 2x and Ultrafast at 6x, the latter only on GPT-6 Astra ($60 / $300) and GPT-6.1 Sol; Claude Opus 5.5 fast mode $8 / $40 | | Long context | Surcharge above a threshold, by vendor | OpenAI (GPT-5.5 and later): above 272K, the whole request at 2x input and cache and 1.5x output; Gemini 3.1 Pro $4 / $18 above 200K; Claude 4.6 and later none, except Haiku 5.5 (5x above 100K) | | Data residency | Premium to keep inference in a region | OpenAI +10% on models since March 5, 2026; GPT-6 Astra on Microsoft Foundry: US Data Zone +10%, no EU Data Zone announced at launch; Anthropic US-only 1.1x; Bedrock and Vertex regional endpoints +10% | | Media and tools | Per image, per second or per call | Nano Banana 2.1 $0.0336 per 1K image; Veo 3.1, now only on Google Cloud, $0.40 per second; Gemini Omni Flash, its replacement on the Gemini API, about $0.10 per second at 720p; Claude web search $10 per 1,000 | ## Prices are splitting, not just falling The trend depends on the tier you buy. 2026 created a premium tier at $10 / $50: Claude Fable 5 (June 9), Fable 5.1 (September 1) and GPT-6 Astra (September 3). The middle got cheaper: GPT-6 Sol ($2 / $10) and Luna ($0.10 / $0.50) launched on September 22 at half or less of their GPT-5.6 predecessors' promotional prices; GPT-6.1 Sol followed at the end of the month at the same price, and GPT-6.1 Terra arrived on October 26 at $1 / $5. The low end too: Claude Haiku 5.5 (October 7) costs $0.10 / $0.50 for prompts up to 100K tokens, 90% less than Haiku 4.5 ($1 / $5), and five times more above that threshold. Per unit of capability, the fall is steep. Epoch AI found the price of reaching a fixed performance level dropped 9x to 900x a year depending on the threshold, with the fastest drops the least certain to last. On ARC Prize data, Claude Opus 4.6 scored 64.6% on ARC-AGI-2 for $2.25 per task in February 2026; DeepSeek V4 Flash scored 61.4% for $0.042 in July. Token counts move too: Anthropic's tokenizer, used since Claude Opus 4.7, produces about 30% more tokens for the same text, so a lower price per token can still mean a higher bill. Haiku 5.5 is the latest case: it counts about 30% more tokens than Haiku 4.5 for the same text, so its 90% cut per token does not translate into the same saving per task. ## Cost per solved task, not price per token The unit that matters is the finished job. A cheaper token that needs two retries is not cheaper. Our working formula: cost per solved task = spend on a route ÷ tasks that pass your evaluation. It captures retries, reasoning and tool calls that a price sheet ignores. Effort shows how far the two diverge. On ARC Prize's ARC-AGI-2 board on September 26, Claude Opus 5.5 cost $0.241 per task at low effort (70.1%), $0.408 at high (93.3%) and $1.85 at max (91.7%): max cost 7.7 times low and 4.5 times high, for a lower score than high. The same holds across models. On version 5.0 of the Artificial Analysis index (October 27) at max effort, GPT-6 Astra costs $6.24 per task against $10.04 for Opus 5.5, even though its tokens cost 2.5 times more: to run the index, Opus 5.5 generated 220 million tokens and Astra 54 million. Treat any cost-per-task figure you haven't measured with suspicion: verify it with your own evals and tune effort as explained in when paying for thinking pays off. ## The levers, in the order we apply them Quality-neutral levers first; trade-offs last, and only with a before-and-after eval. | Lever | Documented effect | Watch out for | | --- | --- | --- | | 1. Caching | Reads at 10% of input (5% on Opus 5.5, Sonnet 5.5 and GPT-6.1 Sol; 2.5% on Fable 5.1); in a 2026 case study, a 99.3% cache hit rate cut API spend 88.6% | Needs a stable prefix; on Anthropic, changing effort invalidates the cache | | 2. Prompt and tool hygiene | Anthropic cut an agent workflow from 150,000 to 2,000 tokens by calling MCP tools from code | Tool definitions are billed: Claude's computer-use toolset adds about 4,500 tokens per request | | 3. Output length | OpenAI's verbosity setting gave 560, 849 or 1,288 tokens on one task | On Opus 5, effort doesn't reliably shorten answers: ask for length in the prompt | | 4. Batch | 50% off at the three big vendors; stacks with caching on Anthropic | Asynchronous; not combinable with Anthropic's fast mode | | 5. Effort | Opus 5.5 on ARC-AGI-2 (September 26): $0.241 per task at low, $1.85 at max | Less effort can cost quality; set it per route | | 6. Routing and cascades | RouteLLM cut cost over 85% on MT Bench keeping 95% of GPT-4's quality; Cursor says its router beats Opus 4.8 at 41% lower cost | Routers need their own evals and a fallback | | 7. Smaller models | GPT-6 Luna costs a twentieth of GPT-6.1 Sol, as Claude Haiku 5.5 does against Sonnet 5.5 for prompts up to 100K tokens; gpt-oss-120b is $0.15 / $0.60 on Groq and Together | Gaps on hard tasks and in Spanish or French | | 8. Distillation | DeepSeek says its 8B distilled model matches Qwen3-235B-thinking on AIME 2024 | Needs data, training and evals; Anthropic blocks extraction of Opus 5.5's reasoning | ## Costs that don't show on the token bill Budget them from day one. - Evals and observability. Every model or prompt change needs a re-run; our evals guide puts a full run at $16 to $32 in its example. - Engineering time. In the same 2026 case study, the local open-weight setup had a fix-commit ratio of 74.9% against 45.9% with the API model. - Retries, refusals and guardrails. Failed attempts are billed, some Fable 5 refusals have been billed since September 24, 2026, and each guardrail is another model call. - Runtime fees. Claude Managed Agents bills $0.08 per session-hour and web search $10 per 1,000 searches. - Lock-in. A US export-control directive forced Anthropic to pull Fable 5 for all users from June 12 to July 1, 2026. - Scheduled price changes. Gemini 3.6 Flash and 3.8 Flash double to $1.50 / $7.50 on January 1, 2027; GPT-5.6 Sol's $4 / $20 is promotional through at least November 21, 2026; and Gemini 4 Argon moves from its $2 / $10 introductory price to $4 / $20 on December 1, 2026. ## Local or API: a break-even method It depends on what you compare against and how busy the machine is. Monthly local cost = hardware ÷ months of life + power + operations time. Break-even = monthly local cost ÷ API cost per request. Then check that the machine can serve that volume and that the local model passes the same eval. An illustrative example, not a quote: a workstation with an NVIDIA RTX 5090 (launch price $1,999, 575 W) running gpt-oss-20b, which LMSYS measured on that card at 8,519 tokens per second reading and 205 writing. Our assumptions: $1,000 for the rest of the machine, 36 months of life, 0.7 kW, 8 hours a day for 20 days a month at $0.15 per kWh (as in Pan et al.), 8 hours of operations a month at $50 an hour, and requests of 2,000 input and 500 output tokens. | API compared | Cost per request | Requests a month to match $500 | Fits local capacity? | | --- | --- | --- | --- | | Claude Opus 5.5 ($4 / $20) | $0.018 | About 27,800 | Yes | | GPT-6.1 Sol ($2 / $10) | $0.009 | About 55,600 | Yes | | GPT-6 Luna ($0.10 / $0.50) | $0.00045 | About 1.1 million | No | | gpt-oss-20b on Groq ($0.075 / $0.30) | $0.0003 | About 1.7 million | No | Against a frontier API, the workstation pays for itself at tens of thousands of requests a month, if a 20B model passes your evals. Against the same open model served by an API, it doesn't at this utilization. Operations is the biggest line: without it, break-even against GPT-6.1 Sol falls to about 11,100 requests. Studies agree on the shape. Pan et al. (2025) see on-premise paying off mainly above 50 million tokens a month or under residency mandates, and the 2026 case study found cached API calls at about $0.57 per million tokens, below the $2.83 of a shared on-premise GPU slice. Epoch AI adds that AI-chip performance per dollar rises about 49% a year: owned hardware ages fast. Before you buyRedo this calculation with your volumes, eval pass rates and labor costs. Compliance can justify local inference on its own; cost alone rarely does below high, steady volume. Options in our local AI hardware guide. ## FinOps for AI: budgets, alerts and reviews Cost control is a routine, not a project. - Tag every call with route, feature, customer, model, effort and cache hit. - Dashboards per route: cost per solved task, cache hit rate, output tokens per task, retries. - Alerts and quotas: daily spend against a baseline, hard caps per key and per team. Agents burn budgets fast: OpenAI says its median researcher's daily token use, at API prices, exceeds $600. - Effort discipline: set it per route, not per request; on Anthropic a change invalidates the cache. - A monthly review of model mix, eval pass rates, price changes and retirement dates, with one owner who can say no. ## Procurement notes for Colombia and Europe Currency, residency and exit terms matter as much as the rate card. Currency. Rate cards are in US dollars, so a budget in pesos or euros adds exchange-rate risk to usage risk. Budget in USD with a buffer, review monthly and have your tax advisor confirm how foreign digital services are taxed in your case; we don't model taxes here. Residency. Keeping inference in a region costs more: +10% for OpenAI regional processing and for Bedrock or Vertex regional endpoints, 1.1x for Anthropic's US-only inference. Buying Claude through AWS, Google Cloud or Microsoft lets you use existing cloud commitments. Retirements and exit. Anthropic publishes minimum availability dates (Opus 5.5 not before September 22, 2027). Haiku 4.5's commitment ran only to October 15, 2026: it is still available as a legacy model, with no retirement date announced, while Sonnet 4.5 retires on November 30, 2026. Negotiate notice periods for retirements and price changes, migration support and data-retention terms, and keep a tested fallback from a second vendor. We help with this in AI consulting. Engineering cost. Tokens are one line of the budget; the team that builds and runs the system is the other. US companies can keep that line predictable, on their own business hours, with nearshore software development in Colombia. ### Key takeaways - Price per token is only the base: caching, reasoning billed as output, batch and surcharges for long context, speed or residency change the bill. - 2026 split the market: a $10 / $50 premium tier appeared while mid-tier prices fell, and price per capability keeps dropping. - Measure cost per solved task: on ARC-AGI-2 (September 26), Opus 5.5 at max effort cost 7.7 times low and scored below high. - Apply caching and batch first; effort, routing, smaller models and distillation trade quality for cost and need a before-and-after eval. - Local beats the API only at high, steady volume or for compliance; operations time is usually the biggest line. ## Sources - Pricing · Anthropic documentation, 2026-10-27 - Effort · Anthropic documentation, 2026-09 - Models overview · Anthropic documentation, 2026-10-27 - Model deprecations · Anthropic documentation, 2026-10-27 - Claude Haiku 5.5: overview and pricing · Anthropic documentation, 2026-10-07 - Introducing Claude Haiku 5.5 · Anthropic, 2026-10-07 - Introducing Claude Sonnet 5.5 · Anthropic, 2026-09-28 - API pricing · OpenAI, 2026-10-27 - GPT-6.1 Sol: model page · OpenAI API documentation, 2026-10-27 - OpenAI launches GPT-6.1 Sol, says it nearly matches GPT-6 Astra and costs less · TechCrunch, 2026-09-29 - Introducing GPT-6.1 Terra · OpenAI, 2026-10-26 - Introducing GPT-6 Sol and GPT-6 Luna · OpenAI, 2026-09-22 - GPT-6 Astra: frontier intelligence for work, now generally available in Microsoft Foundry · Microsoft Azure, 2026-09-03 - Gemini Developer API pricing · Google AI for Developers, 2026-10-27 - Gemini API deprecations · Google AI for Developers, 2026-10-22 - Gemini API changelog: Nano Banana 2.1 generally available · Google AI for Developers, 2026-10-06 - Generate videos with Veo 3.1 (preview shut down on October 22, 2026) · Google AI for Developers, 2026-10-22 - Generative AI pricing (Veo 3.1 and Gemini Omni Flash) · Google Cloud, 2026-10-27 - Gemini 4 Argon · Google, 2026-09-30 - ARC-AGI leaderboard · ARC Prize, 2026-09-26 - AI model leaderboards (Intelligence Index v5.0) · Artificial Analysis, 2026-10-27 - Intelligence Index methodology and version history · Artificial Analysis, 2026-10-27 - Claude Opus 5.5: intelligence, performance and price · Artificial Analysis, 2026-10-27 - GPT-6 Astra: intelligence, performance and price · Artificial Analysis, 2026-10-27 - LLM inference prices have fallen rapidly but unequally across tasks · Epoch AI, 2025-03-12 - RouteLLM: An Open-Source Framework for Cost-Effective LLM Routing · LMSYS, 2024-07-01 - Code execution with MCP · Anthropic, 2025-11-04 - A Cost-Benefit Analysis of On-Premise Large Language Model Deployment: Breaking Even with Commercial LLM Services · Pan et al., arXiv, 2025-08-30 - GeForce RTX 5090 · NVIDIA, 2025-01-06 Editorial note: this analysis reflects the public information available on the review date. Models, prices and rules change fast; every third-party figure links to its source, and our opinions are labeled as such. Spotted an error? Write to contact@slash-digital.io. Frequently asked questions ## The questions we hear often **Why is output more expensive than input?** Vendors charge more for generating than for reading: on current models output costs five to six times input (Claude Opus 5.5 $4 / $20, GPT-6.1 Sol $2 / $10, Gemini 3.8 Flash $0.75 / $3.75). OpenAI and Google bill reasoning tokens as output, so effort settings weigh on the bill. **How much does prompt caching save?** Cache reads cost 10% of the input price at Anthropic, on OpenAI's GPT-5 family, GPT-6 Sol, GPT-6 Astra and GPT-6 Luna and on Gemini 3.8 Flash, 5% on Claude Opus 5.5 and Sonnet 5.5, GPT-6.1 Sol, GPT-6.1 Terra and Gemini 4 Argon, and 2.5% on Claude Fable 5.1. Writes cost extra (1.25x or 2x at Anthropic), so caching pays when the same prefix is reused while the cache is alive. **Is the Batch API worth it?** For anything that doesn't need an immediate answer, yes: Anthropic, OpenAI and Google charge 50% less for batch jobs, and at Anthropic the discount stacks with caching. **When does local inference beat the API?** At high, steady volume against expensive APIs, or when compliance requires it. In our illustrative example, one RTX 5090 workstation matches GPT-6.1 Sol at about 55,600 requests a month but never the same open model served by an API. **How should a company budget for AI billed in dollars?** Budget in USD with a buffer for the exchange rate, set quotas and alerts per route, review monthly with announced price changes in view, and have your tax advisor confirm how foreign digital services are treated.