GEM-4C 558 -4%
GPT-61S 555 -3.8%
CL-SN55 553 -3.8%
CL-OP55 551 -4%
GPT-61TERR 547 -3.5%
GEM-4CM 546 -4%
CL-OP55EH 542 -3.9%
GPT-61SOLXH 541 -3.6%
GPT-61SOLH 527 -3.7%
CL-SN55XH 523 -3.5%
GEM-4CL 520 -3.9%
GPT-61TERRXH 507 -3.4%
CL-OP55H 506 -4%
GPT-61SOLM 500 -3.5%
KIMI-K31P 499 -3.5%
GEM-4AR 498 -3.9%
GPT-61TERRH 490 -3.2%
CL-OP55M 483 -3.6%
MUSE-WONDFUL 479 -3.8%
CL-SN55H 467 -3.3%
CL-HK55 463 -2.9%
GPT-61TERRM 462 -3.1%
GPT-6A 451 -4.7%
MUSE-WOND 451 -3.6%
CL-HK55XH 450 -3%
GPT-6ASTRXH 440 -4.8%
GROK-47 440 -3.1%
GROK-47H 438 -2.9%
GPT-6ASTRH 437 -4.8%
DSK-V-340615 436 -2.9%
QWEN-38X 428 -2.9%
GPT-61SOLL 423 -3.2%
GLM-53 420 -3%
MIMO-V26P 420 -2.6%
CL-HK55H 419 -2.8%
LING-31F1019 417 -2.6%
GEM-38F 416 -2.8%
GPT-6ASTRM 415 -4.6%
MMAX-M31F 414 -2.8%
GEM-38FM 412 -2.8%
GPT-61TERRNO 411 -2.8%
GROK-47L 410 -2.8%
DSK-V-3E48BD 410 -2.6%
GPT-6L 407 -2.6%
GLM-53F 407 -2.6%
STEP-5 405 -2.6%
CL-FABL51XH 403 -4.7%
GPT-61TERRL 403 -2.9%
MIMO-V26F 401 -2.4%
CL-FABL51 399 -4.5%
CL-SN55M 395 -2.9%
CL-OP55L 393 -3.4%
MIST-LARG4 390 -2.7%
CL-FABL51H 386 -4.5%
CL-HK55M 386 -2.3%
GPT-6LUNAXH 377 -2.6%
MUSE-SPAROPE 375 -2.6%
MIST-LARG4M 371 -2.4%
DSK-V4P 365 -2.4%
GPT-6LUNAH 365 -2.1%
GLM-53L 364 -2.4%
CL-SN55L 362 -2.7%
GEM-38FL 361 -2.4%
GPT-6ASTRL 359 -4.5%
CL-FABL51M 359 -4.5%
QWEN-827B 349 -2.2%
CL-HK55L 347 -2%
MMAX-M3 345 -2%
GPT-6LUNAM 344 -2.3%
DSK-V-02811D 344 -2%
K2-HORI375BA 342 -2%
REFL-BEAM 338 -2%
QWEN-827BM 332 -1.8%
CL-FABL51L 329 -4.4%
QWEN-827BL 320 -1.8%
NEMO-35SUPEV 313 -1.9%
INKL 295 -1.7%
NVID-NEMO3UL 295 -1.3%
GEM-35FLITE 295 -1.3%
GPT-6LUNAL 287 -1.4%
GPT-6LUNANON 286 -1%
NANB-333B 285 -2.7%
QWEN-827BNON 282 -1.4%
COMM-APLUS 266 -1.1%
MUSE-GLIM 263 -1.1%
NEMO-35LIGH 261 -0.8%
GRAN-428B 258 -0.8%
MIST-SMAL4 252 -0.4%
INKL-SM 235 -2.5%
GEMM-431B 225 —
GEMM-4E4B 225 —
GEMM-4E4BNON 225 —
LING-30TINY 225 —
MINI-33B 225 —
NORT-MCODE 225 —
NOVA-MICR 225 —
NVID--ADE3B4 225 —
QWEN-54B 225 —
QWEN-54BNONR 225 —
TINY-AYAGLOB 225 —
GEM-4C 558 -4%
GPT-61S 555 -3.8%
CL-SN55 553 -3.8%
CL-OP55 551 -4%
GPT-61TERR 547 -3.5%
GEM-4CM 546 -4%
CL-OP55EH 542 -3.9%
GPT-61SOLXH 541 -3.6%
GPT-61SOLH 527 -3.7%
CL-SN55XH 523 -3.5%
GEM-4CL 520 -3.9%
GPT-61TERRXH 507 -3.4%
CL-OP55H 506 -4%
GPT-61SOLM 500 -3.5%
KIMI-K31P 499 -3.5%
GEM-4AR 498 -3.9%
GPT-61TERRH 490 -3.2%
CL-OP55M 483 -3.6%
MUSE-WONDFUL 479 -3.8%
CL-SN55H 467 -3.3%
CL-HK55 463 -2.9%
GPT-61TERRM 462 -3.1%
GPT-6A 451 -4.7%
MUSE-WOND 451 -3.6%
CL-HK55XH 450 -3%
GPT-6ASTRXH 440 -4.8%
GROK-47 440 -3.1%
GROK-47H 438 -2.9%
GPT-6ASTRH 437 -4.8%
DSK-V-340615 436 -2.9%
QWEN-38X 428 -2.9%
GPT-61SOLL 423 -3.2%
GLM-53 420 -3%
MIMO-V26P 420 -2.6%
CL-HK55H 419 -2.8%
LING-31F1019 417 -2.6%
GEM-38F 416 -2.8%
GPT-6ASTRM 415 -4.6%
MMAX-M31F 414 -2.8%
GEM-38FM 412 -2.8%
GPT-61TERRNO 411 -2.8%
GROK-47L 410 -2.8%
DSK-V-3E48BD 410 -2.6%
GPT-6L 407 -2.6%
GLM-53F 407 -2.6%
STEP-5 405 -2.6%
CL-FABL51XH 403 -4.7%
GPT-61TERRL 403 -2.9%
MIMO-V26F 401 -2.4%
CL-FABL51 399 -4.5%
CL-SN55M 395 -2.9%
CL-OP55L 393 -3.4%
MIST-LARG4 390 -2.7%
CL-FABL51H 386 -4.5%
CL-HK55M 386 -2.3%
GPT-6LUNAXH 377 -2.6%
MUSE-SPAROPE 375 -2.6%
MIST-LARG4M 371 -2.4%
DSK-V4P 365 -2.4%
GPT-6LUNAH 365 -2.1%
GLM-53L 364 -2.4%
CL-SN55L 362 -2.7%
GEM-38FL 361 -2.4%
GPT-6ASTRL 359 -4.5%
CL-FABL51M 359 -4.5%
QWEN-827B 349 -2.2%
CL-HK55L 347 -2%
MMAX-M3 345 -2%
GPT-6LUNAM 344 -2.3%
DSK-V-02811D 344 -2%
K2-HORI375BA 342 -2%
REFL-BEAM 338 -2%
QWEN-827BM 332 -1.8%
CL-FABL51L 329 -4.4%
QWEN-827BL 320 -1.8%
NEMO-35SUPEV 313 -1.9%
INKL 295 -1.7%
NVID-NEMO3UL 295 -1.3%
GEM-35FLITE 295 -1.3%
GPT-6LUNAL 287 -1.4%
GPT-6LUNANON 286 -1%
NANB-333B 285 -2.7%
QWEN-827BNON 282 -1.4%
COMM-APLUS 266 -1.1%
MUSE-GLIM 263 -1.1%
NEMO-35LIGH 261 -0.8%
GRAN-428B 258 -0.8%
MIST-SMAL4 252 -0.4%
INKL-SM 235 -2.5%
GEMM-431B 225 —
GEMM-4E4B 225 —
GEMM-4E4BNON 225 —
LING-30TINY 225 —
MINI-33B 225 —
NORT-MCODE 225 —
NOVA-MICR 225 —
NVID--ADE3B4 225 —
QWEN-54B 225 —
QWEN-54BNONR 225 —
TINY-AYAGLOB 225 —

Delayed ticker data. Last calculated 5 Nov 2026, 11:04 pm Sydney time.

← Back to feed

OpenRouter Cost Audit: Billed Rates for 20 Image Models Diverge From Listed Prices by Up to 50%

In September 2026, OpenRouter ran a systematic cost audit of 20 commercially available image generation models, billing each through its API and recording what it actually charged rather than what was listed. The results expose a market where developer cost estimates routinely break against actual invoices.

What They Were Actually Billed

GPT-image-2 returned a 1024x1024 PNG for $0.006 — the lowest billed cost across all tested models. At the other end, Google’s Gemini 3 Pro Image billed $0.1344 for the same resolution, a 22x gap at the extremes.

ModelBilledOutputBilling Unit
OpenAI GPT-image-2$0.00601024x1024 PNGtoken
FLUX.2-Klein-4B$0.01401024x1024 JPEGmegapixel
Riverflow v2.5-Fast$0.01761024x1024 WebPimage
FLUX.2-Pro$0.03001024x1024 JPEGmegapixel
Qwen Image 3$0.03001024x1024 PNGimage
Krea 2 Medium$0.03001024x1024 PNGnot published
GPT-image-1-mini$0.03331024x1024 PNGtoken
Gemini 3.1 Flash Lite Image$0.03361024x1024 JPEGtoken
Recraft v4.1$0.03501024x1024 WebPimage
Seedream 5.0 Lite$0.03502048x2048 JPEGimage
MAI Image 2.5$0.04821024x1024 PNGtoken
FLUX.2-Flex$0.05001024x1024 JPEGmegapixel
Grok Imagine Quality$0.05001024x1024 JPEGimage
Grok Imagine 2.0$0.06001024x1024 JPEGimage
Riverflow v2.5-Pro$0.06411024x1024 WebPimage
Gemini 3.1 Flash Image$0.06721024x1024 PNGtoken
FLUX.2-Max$0.07001024x1024 JPEGmegapixel
Recraft v4.1 Vector$0.0800vector SVGimage
Seedream 5.0 Pro$0.09002048x2048 JPEGimage
Gemini 3 Pro Image$0.13441024x1024 PNGtoken

The Listed-vs-Billed Gap

The audit’s headline finding is not the absolute prices but the divergence between what providers list and what they charge. Sourceful’s Riverflow v2.5-Pro listed $0.13 per image but billed $0.064 — exactly half. Black Forest Labs’ FLUX.2-Flex listed $0.06 per megapixel and billed $0.05. FLUX.2-Max, by contrast, listed $0.07 and billed $0.07 exactly.

Krea 2 Medium presented a different problem: no published pricing at all, but billed at $0.030 — in line with mid-table competitors, just undocumented before the transaction.

The discounts appear to come from caching, provider-level bulk agreements, and routing optimization rather than advertised rate cuts. For developers building cost models, the implication is that listed rates are a ceiling, not a floor, and actual spend may be substantially lower.

Billing Unit Diversity as a Trap

The audit surfaces a structural issue for developers estimating at scale: billing units are not standardized. Token-based models (OpenAI, Google, Microsoft) price based on image resolution — a switch from 1024x1024 to higher output will increase the bill proportionally. Megapixel billing (Black Forest Labs) normalizes to area in a similar way. Fixed-image billing (most other providers) charges the same regardless of output resolution at the tested size.

The practical trap: comparing GPT-image-2 at $0.006 against FLUX.2-Pro at $0.030 on a fixed-size request understates the gap when factoring in 2K or 4K outputs, where token-based billing scales up and fixed-image billing stays flat.

Resolution Arbitrage

ByteDance Seedream’s pricing strategy stands out. Seedream 5.0 Lite returned a 2048x2048 image — four times the pixel area of a standard 1024x1024 request — for $0.035, the same price as Recraft v4.1 at standard resolution. Seedream 5.0 Pro returned the same 2048x2048 output for $0.09. On a per-pixel basis, both Seedream tiers offer the highest resolution value in the tested set.

Recraft v4.1-Vector billed $0.08 for an SVG output — a format that scales to any resolution without quality loss — putting it in a different comparison class than raster-output competitors.

Bottom Line

The cheapest route for standard 1024x1024 raster output is GPT-image-2 at $0.006. FLUX.2-Klein-4B is the cheapest open-weight alternative at $0.014. For high-resolution needs, Seedream 5.0 Lite’s 2048x2048 at $0.035 is structurally cheaper than any token-billed model at equivalent output.

Listed prices should be treated as approximate. The audit found consistent discounts, with the largest divergence reaching 50% below the listed rate. Budget estimates built on provider marketing pages will overstate actual API spend.