FLUX API Pricing Comparison 2026: FLUX.2 Klein, Pro, Flex, and Max Cost Per Image

FLUX API pricing, as of 2026-07-07, runs from about $0.014 per image at the low end (FLUX.2 Klein 4B) to $0.07 per image at the high end (FLUX.2 Max), based on Black Forest Labs' own pricing docs and fal.ai's per-megapixel rate card. At 1,000 images that is roughly $14 for Klein 4B versus $70 for Max, a 5x spread driven mostly by output quality tier, not resolution alone. This article puts those numbers side by side across the official BFL page, fal.ai, and TokenLab's live model directory, so you can pick a tier before writing integration code.

Key Takeaways

Source Snapshot

Source	What it confirms	Observed
Black Forest Labs pricing docs (docs.bfl.ml/quick_start/pricing)	Official per-credit and per-image rates for FLUX.2 Klein 4B/9B, Pro, Flex, Max	2026-07-07
fal.ai FLUX.2 model page (fal.ai/flux-2)	Per-megapixel pricing for FLUX.2 dev, pro, flex, max	2026-07-07
TokenLab model directory (tokenlab.sh/en/models)	Live per-image lock pricing across 17 tracked FLUX variants, both FLUX.1 legacy and FLUX.2 current	2026-07-07
Replicate pricing (replicate.com/pricing)	General compute pricing page; verify FLUX-specific per-image rates directly before quoting them, since this packet does not confirm exact Replicate FLUX numbers	2026-07-07

FLUX API Pricing by Variant and Provider

This table pulls only confirmed figures. Rows without a confirmed number are left blank rather than estimated.

FLUX Variant	BFL Official (direct)	fal.ai	TokenLab Platform	Est. cost per 1,000 images
FLUX.2 Klein 4B	$0.014/image	-	$0.014/image	~$14
FLUX.2 Klein 9B	$0.015/image	-	$0.015/image	~$15
FLUX.2 dev	-	$0.012/MP	-	~$12 (per MP, verify at output size you need)
FLUX.1 [dev] (legacy)	-	-	$0.025/image	~$25
FLUX.2 Pro (text-to-image)	$0.03/image	$0.03/MP	-	~$30
FLUX.1.1 Pro (legacy)	-	-	$0.04/image	~$40
FLUX.2 Pro (image editing)	$0.045/image	-	-	~$45
FLUX.2 Flex	$0.05/image	$0.05/MP	$0.05/image	~$50
FLUX Pro 1.1 Ultra (legacy)	-	-	$0.06/image	~$60
FLUX.2 Max	$0.07/image	$0.07/MP	$0.07/image	~$70

Note on fal.ai's per-MP rates: cost scales with output resolution, so a 1 megapixel image at the listed rate matches the number above, but larger outputs (2MP, 4MP) will cost proportionally more. Confirm the exact output size your product requests before treating fal.ai's per-MP price as a flat per-image cost.

One consistency check worth noting: BFL's official Flex ($0.05) and Max ($0.07) rates match both fal.ai's per-MP starting price and TokenLab's platform lock price exactly. That agreement across three independent pricing surfaces is a good signal these are stable current-generation numbers rather than a promotional rate that will move soon.

FLUX.2 vs FLUX.1: What Changed

Black Forest Labs' current pricing docs are built around FLUX.2 naming (Klein 4B, Klein 9B, Pro, Flex, Max) rather than the older FLUX.1 [schnell]/[dev]/[pro]/[pro Ultra] scheme. FLUX.2 uses megapixel-based pricing that scales with output resolution, and BFL confirms fine-tuned FLUX.2 public beta endpoints bill at the same rate as their base endpoint, so a custom fine-tune of Flex still costs $0.05/image rather than a premium rate.

TokenLab's directory still lists legacy FLUX.1 variants (flux-1-dev, flux-1.1-pro, flux-pro-1.1-ultra) alongside the current FLUX.2 line, because existing integrations built against those model IDs are still live. If you are starting a new integration today, default to FLUX.2 Klein or Pro rather than the FLUX.1 equivalents: same or lower cost, current generation, and less likely to hit a deprecation notice in the next few quarters.

How FLUX Pricing Models Work

FLUX pricing is built on a credit system where BFL's own docs state 1 credit equals $0.01 USD. Cost scales along three axes:

• Variant tier - Klein is the budget tier, Pro sits in the middle, Flex and Max are the premium tiers. This is the biggest driver of your bill, roughly a 5x spread from cheapest to most expensive confirmed rate.
• Output resolution - FLUX.2's megapixel-based pricing (confirmed on both BFL and fal.ai) means a 2MP output costs roughly double a 1MP output at the same variant tier. Check your default output size in your integration config, since many SDKs default to a higher resolution than you assume.
• Task type - BFL's docs distinguish text-to-image from image editing at the Pro tier ($0.03 vs $0.045), a 50 percent premium for editing calls. If your workflow does both, model each separately rather than using a single blended rate.

Comparing FLUX Variants: What Affects Your Bill Beyond the Sticker Price

Factor	Impact on cost
Model variant (Klein vs Pro vs Flex vs Max)	Largest single driver; confirmed spread is $0.014 to $0.07 per image
Output resolution (megapixels)	FLUX.2's per-MP pricing means larger outputs cost proportionally more
Task type (generation vs editing)	BFL's Pro tier charges $0.045 for editing vs $0.03 for text-to-image, a 50% premium
Retry/regeneration rate	Failed or rejected generations still consume credits on most platforms
Fine-tuned endpoints	BFL bills fine-tuned FLUX.2 public beta endpoints at the same rate as base, no premium confirmed
Legacy vs current generation	FLUX.1 legacy variants on TokenLab range $0.025 to $0.06/image, overlapping with but not identical to FLUX.2 pricing

If you run a product where users generate multiple variations per prompt, common in creative tools, your effective cost per accepted image can be 2 to 4x the advertised per-generation price once retries are factored in. Model this against the table above before committing to a tier, since a $0.014 Klein image that needs three retries to get an acceptable result costs more in practice than a $0.03 Pro image that succeeds on the first try.

Direct API vs Aggregator: Which Fits Your Workflow

Developers generally choose between two integration paths:

1. Direct provider API (Black Forest Labs' own endpoint, confirmed pricing at docs.bfl.ml/quick_start/pricing, or fal.ai's hosted FLUX.2 models at fal.ai/flux-2). This gives the clearest line to the provider's published rate and the least abstraction between your code and the billing meter.
2. Aggregator platforms that route requests across multiple image models, including FLUX.2 variants, through one API key and one billing dashboard. TokenLab's model directory lists FLUX.2 Klein, Pro, Flex, and Max alongside the legacy FLUX.1 line and other image models with matching per-image lock pricing (tokenlab.sh/en/models, observed 2026-07-07), useful for side-by-side evaluation before you commit budget to a single provider.

The tradeoff is integration overhead versus flexibility. If you expect to switch between FLUX, Nano Banana style models, or other image APIs based on cost or quality testing, an aggregator reduces the engineering cost of maintaining multiple SDKs and auth flows. For a broader look at how model aggregation affects pricing and routing decisions across text and image models, see the OpenRouter comparison.

Checklist: Evaluating FLUX API Pricing for Your Project

Use this before selecting a provider or variant:

• Confirm current per-image or per-MP pricing on BFL's or fal.ai's official page directly; rates for newer tiers can shift as infrastructure changes.
• Test FLUX.2 Klein first if your use case tolerates lower fidelity at $0.014-$0.015/image, then benchmark Pro ($0.03), Flex ($0.05), and Max ($0.07) against your actual prompts.
• Calculate expected retry rate from a pilot batch of 100-200 images, not just the advertised per-image cost.
• Check whether image editing calls are billed at a premium over text-to-image, since BFL confirms a 50% uplift at the Pro tier.
• If migrating from FLUX.1 legacy variants, compare against FLUX.2 equivalents since Klein and Flex/Max pricing is competitive or lower.
• Verify rate limits and concurrency caps, which affect throughput-dependent applications more than raw per-image cost.
• Confirm whether fine-tuned endpoints carry a pricing premium; BFL's current docs state fine-tuned FLUX.2 public beta bills at base rate.

If your product also needs video generation alongside static images, review current options and confirmed per-second pricing in best AI video models API 2026, since providers like PixVerse and Veo bill per second rather than per image.

FLUX vs Other Image Models: Where It Fits

FLUX is one option among several production-grade image generation APIs. Developers comparing FLUX.2 against alternatives should weigh:

• Prompt adherence - FLUX variants are generally noted for strong prompt-following behavior, which matters for programmatic generation where you can't manually adjust prompts per output.
• Pricing structure - FLUX.2's per-megapixel model differs from flat per-image pricing used by some competing image APIs; confirm which model your budget math assumes.
• Cost at scale - a $0.014 Klein image that looks competitive at low volume can shift ranking once you're generating thousands of images daily, especially with retries factored in. At 100,000 images, the gap between Klein ($1,400) and Max ($7,000) is $5,600, worth modeling before you scale a pilot.

For a full breakdown of image model options and their pricing structures, see best AI image models API 2026. To see how FLUX pricing stacks up against a wider set of models across categories, the pricing comparison page aggregates current rates in one view.

Compare image models to see current FLUX.2 variant pricing alongside other image generation APIs before you commit to a single provider.

Making the Final Call

There's no single cheapest FLUX tier that fits every product. A high-volume batch job with tolerant quality requirements should default to FLUX.2 Klein at $0.014-$0.015/image. A client-facing product where every generated image is shown to an end user typically justifies Pro, Flex, or Max pricing ($0.03 to $0.07/image) despite the higher per-image cost, because regeneration and support costs from low-quality output often exceed the price difference.

Run a pilot: generate 100-200 images across two or three variants using your actual prompts, track acceptance rate, and calculate effective cost per accepted image using the per-1,000-image figures above as your baseline. This number, not the headline per-image price, is what should drive your final provider selection.

Compare live FLUX pricing and variant specs in the TokenLab model directory before you commit to a production integration.

FAQ

How much does FLUX API pricing cost through an API? Confirmed 2026-07-07: FLUX.2 Klein 4B and Klein 9B start at $0.014 and $0.015 per image on Black Forest Labs' own pricing page. FLUX.2 Pro starts at $0.03/image for text-to-image and $0.045/image for editing. FLUX.2 Flex is $0.05/image and FLUX.2 Max is $0.07/image, confirmed identically on fal.ai and TokenLab's platform. Legacy FLUX.1 variants (dev, 1.1 Pro, Pro 1.1 Ultra) still run $0.025 to $0.06/image on TokenLab if you have an existing integration.

When should developers use FLUX API pricing instead of a direct provider account? Use a direct BFL or fal.ai account when you need the clearest line to the official rate and don't expect to switch models. Use an aggregator like TokenLab when you want to test FLUX.2 against other image models under one API key and billing dashboard, since the per-image lock prices match BFL's official numbers exactly for Klein, Flex, and Max.

How does TokenLab help compare FLUX API pricing with related models? TokenLab's model directory lists FLUX.2 and legacy FLUX.1 variants alongside other image APIs with matching per-image pricing metadata in one place, so you can benchmark cost per accepted image across models without maintaining separate SDK integrations for each provider before committing budget.